government-ai-worldb093806d

AI IN HEALTHCARE: A Broad Overview of the International State of Play

Sat, 14 Mar 2026 07:07:03 GMT

From Surgical Robots to Data Breaches:
What the Numbers Tell Us About AI in Healthcare Right Now

Artificial Intelligence in healthcare is, arguably, the fastest-growing, most opportunity-laden - and also the most risk-ridden - field of AI application globally right now . . . and almost certainly will be for years to come.

According to a February 23, 2026, analysis by Fortune Business Insights of AI in the global healthcare sector, the 2025 market was valued at USD 39.34 billion – with growth projected to reach USD 56.01 billion in 2026, and USD 1,033.27 billion by 2034.

In 2025, the United States dominated the AI-in-healthcare market with a market share of 44.5 percent. That market is propelled by the high adoption of AI-based technologies in medical imaging, diagnostics, and patient monitoring. Its growth is also fueled by major collaborations between recognised top-tier players.

Other countries are, however, moving along as fast as they can.

Japan is making increasing investments in AI-based medical devices for the diagnosis of various diseases, along with its general focus on integrating advanced technology into its healthcare system.

China's market is expanding due to significant investments driven, as you would expect, by the push to upgrade its healthcare infrastructure to handle the needs of its large population.

Europe appears to be concentrating on increasing AI integration into hospitals and clinics for managing patient health records, administrative workflows, diagnostics, and surgery. The presence of major pharma and biotech companies adopting AI for innovative drug discovery, particularly in the UK and Germany, is a key growth factor.

The opportunities ar e truly futuristic-sounding albeit they’re already here: robot-assisted surgery being a prime example, with that specific application capturing the largest market share in 2024 and being driven by the rising prevalence of chronic diseases and a growing demand for minimally invasive surgical options that offer quicker patient recovery and better outcomes.

AI in Mental Health Disorder Diagnosis & Management: A Prominent Market Opportunity

The rising, global prevalence of mental health disorders, such as depression, anxiety, and stress-related conditions – attributed to fast-paced lifestyles, social disconnection, and economic uncertainties – is leading to an increased demand for advanced solutions in this sector.

Mental Health America Inc’s ‘Prevalence of Mental Illness 2024’ report estimated that 60 million Americans experience some degree or form of mental illness, with 5.8 percent of the population experiencing it to a severe degree.

The social stigma, discrimination and other personal impacts associated with mental health issues, however, and the resultant reticence of people to seek help, places a significant gap between patients and providers . . . offering both challenges and opportunities for existing providers and future developers of AI-based solutions for handling sensitive conditions.

In April 2024, NextGen Healthcare tackled the shortage of mental and behavioural health professionals with the market introduction of NextGen Ambient Assist. This AI-powered solution transcribes patient/provider conversations in real-time, with the ability to summarise encounters within 60 seconds. It’s promoted as saving providers up to two hours of documentation time daily.

Everybody’s After A Slice of the R&D Funding Pie

Implementing artificial intelligence-based solutions in healthcare often requires significant financial investment. This encompasses the costs of cutting-edge technology, hiring qualified personnel, and maintaining infrastructure. Consequently, the high expenses associated with implementation pose challenges for organisations seeking to integrate AI into their daily operations.

This highlights the need for continuous improvement through Research and Development.

On which note, increasing investments by governments in R&D, and substantial funds by private enterprises in AI technologies tailored for healthcare, is one of the prominent drivers of the market.

Global AI healthcare R&D investment is certainly accelerating. In 2025, AI-enabled healthcare start-ups captured 62 percent of all digital health venture funding in the United States, raising an average of USD 34.4 million per round – an 83 percent premium over non-AI start-ups.

Six out of 11 new AI unicorns created in Q1 2025 were healthcare companies.

Governments are also active investors: the EU's Digital Europe and Horizon programs are funding clinical AI pilots, testing facilities and cross-border data infrastructure. In the U.S., the National Institutes of Health and DARPA have ongoing AI in healthcare research programs, while the UK's MHRA launched a National Commission on the Regulation of AI in Healthcare in September 2025 specifically to shape the next generation of investment and governance.

Notwithstanding all this R&D investment, though, the gap between what is being built and what has been rigorously proven to work – safely, at scale, and in real clinical settings – remains significant and, in some cases, dangerously wide.

The Downside Is Real . . . and Under-Played

What’s found in industry and trade press out there right now is a predominantly very positive framing of AI’s application and benefits. Analyses, however, appear less thorough in its analysis of the downsides, risks, and other aspects of Artificial Intelligence that have – and remain – less than adequately mitigated.

These risks aren’t minimal. Neither are their impacts. And neither are the number of stakeholder groups those impacts hit. Data breaches are the highest profile and most en masse risk categories.

Data breaches come in many shapes and sizes, but basically involve unauthorised access, disclosure, or theft of sensitive personal and medical information. Hardly minimal in impact for those whose data has been stolen.

These breaches can occur through various channels, including cyber-attacks, insider threats, and accidental data exposure. From a sector and governmental perspective, it leads – as a minimum - to declining patient trust in general, and an emboldening of cybercrime perpetrators.

From a market perspective, where health care data compromises occur, it fosters mistrust between patients and providers, impeding market growth.

Data Breach Examples

The healthcare sector in a variety of countries has seen major data privacy breaches in the last year alone.

New Zealand provides two such examples:

In late December 2025, ransom hackers accessed the "My Health Documents" module of Manage My Health – the country's largest patient-facing health portal, with approximately 1.8 million registered users. Not a small number for a country with a population of only 5.3 million people.

Medical records, GP referral letters, hospital discharge summaries and clinical correspondence belonging to between 120,000 and 127,000 patients were accessed and downloaded, with the hackers threatening to release the data on the dark web unless a ransom was paid. The breach primarily affected 45 general practices in Northland and 355 referral-originating practices across multiple regions.

Health Minister Simeon Brown described it as "incredibly concerning" and commissioned an urgent Ministry of Health review. The Office of the Privacy Commissioner noted that it was one of the most serious privacy incidents in New Zealand's history.

A second healthcare provider, CanopyHealth, separately revealed in January 2026 that it had been targeted in a cyberattack in July 2025 – a delay in disclosure that drew sharp criticism from clients and privacy advocates.

On a much larger and global note, in October 2023, 23andMe experienced the data breach of genetic information, resulting in the exposure of the personal information of some 14,000 individuals.

Other notable data and privacy breaches have included the February 2024 ransomware attack on Change Healthcare – a subsidiary of UnitedHealth Group and the United States' largest healthcare payment clearinghouse – which exposed the protected health information of an estimated 192.7 million individuals, making it the largest healthcare data breach in history.

The attack cost Change Healthcare an estimated USD 2.87 billion and forced the company to provide USD 9 billion in no-interest advances to keep healthcare providers solvent after claims systems froze.

In May 2024, Ascension Health – a Catholic health system operating 142 hospitals – fell victim to a Black Basta ransomware attack that caused an electronic health record outage lasting almost four weeks.

According to the HIPAA Journal – the leading provider of HIPAA training, news, regulatory updates, and independent compliance advice – 2025 was “another bad year for healthcare data breaches . . . with almost 57 million individuals known to have been affected . . . and at least 642 data breaches affecting 500 or more individuals.

Lack of AI-Skilled Professionals in Healthcare Sector

The risks of the rapid introduction and integration of Artificial Intelligence into healthcare settings are exacerbated - greatly - by the sector's lack of AI-skilled professionals. That, in turn, is exacerbated further by the reluctance among large swathes of medical practitioners to adopt AI.

Occurrences like large-scale and high-profile medical data breaches don’t help.

Nor does the lack of skill within the existing personnel base of the sector in terms of the operation of AI-based systems. Many educational institutions are struggling to keep pace, resulting in a shortage of graduate programs specialising in AI for healthcare applications.

Additionally, many healthcare professionals are hesitant to adopt AI technologies due to fear of job displacement and/or distrust in the accuracy and reliability of AI-driven tools. It’s a completely understandable reluctance to trust Artificial Intelligence to make clinical decisions that are traditionally the domain of human practitioners.

Notwithstanding, the AI wave in the overarching sense, is unlikely to be stemmed to any great degree. The potential in areas like robotics for surgical precision, for facilitating minimally nvasive procedures versus conventional surgery techniques, and for general operational efficiency, is too strong a driver to hold back progress indefinitely.

Specifically in the surgical context, key medtech companies are collaborating with AI solution providers to incorporate AI into their instruments to aid precision.

In March 2024, Johnson & Johnson Services Inc. partnered with NVIDIA Corporation to accelerate and scale Artificial Intelligence for surgery and to form a "digital surgery ecosystem".

The partnership – formalised through a Memorandum of Understanding announced at the NVIDIA GTC conference in March 2024 – combines J&J's medtech presence in over 80 percent of the world's operating rooms with NVIDIA's IGX edge computing platform and Holoscan edge AI platform.

The stated goals are to enable real-time analysis of surgical data, to broaden the global availability of AI algorithms for surgical decision-making, and to support AI-powered applications across the connected operating room – from pre-operative planning through to post-operative documentation.

In June 2025, the collaboration was extended through the launch o f the Polyphonic AI Fund for Surgery, which added Amazon Web Services as a partner and opened funding to academic institutions, developers, start-ups and researchers working on AI model development, data engineering and AI governance in surgery.

Where to From Here?

For now, though, the Stanford-Harvard report – t he inaugural State of Clinical AI (2026) , released in January 2026 by ARISE, the AI Research and Science Evaluation network, a Stanford-Harvard research collaboration led by clinicians and computer scientists from both institutions — points to AI investment outpacing the evaluation practices needed to justify it . . . “solution” developers’ keenness to secure and expand their slice of this expanding market would suggest that vendors and would-be vendors need to spend more time and investment in the business case stage.

For now, AI adoption is highest for platforms facilitating documentation, clinical workflow orchestration and data infrastructure – areas with measurable administrative ROI – faster than for clinical applications with evidence of patient outcome improvement. In other words, the “boom” is more consistently occurring in applications that reduce paperwork, as opposed to those that save lives.

According to an article in the company news publication of TATEEDA Global, a San Diego-based custom healthcare software development company that has built HIPAA-compliant AI systems for U.S. healthcare organisations since 2013, in 2025, the North American landscape, by way of example, was crowded with systems focusing on the likes of AI-assisted ambient note-taking in exam rooms, AI triage in radiology, smart routing in call centres, and “remote patient monitoring programs that actually change re-admission curves".

The article went on to say that “at the same time, most organisations are still wrestling with ungainly data stacks, fragmented workflows, and early attempts at governance boards and 'AI formularies'.

(For the record, an AI formulary is a curated, approved list of AI tools that a healthcare organisation has formally evaluated and sanctioned for use – modeled directly on the concept of a drug formulary, which lists medications approved for prescribing within a hospital or health system. Just as a drug formulary exists to ensure that only tested, safe and appropriate medications are used on patients, an AI formulary is an attempt to bring the same discipline to AI adoption – preventing clinicians from independently deploying untested or unvetted tools in clinical settings.)

“That tension between proven value and uneven execution,” Tateeda points out, “is exactly what healthcare AI technology trends will need to resolve in 2026: fewer one-off experiments, more shared building blocks for data, models, safety, and monitoring across entire enterprises.”

It says 2026 “is shaping up as the year where scale and discipline matter as much as raw innovation.

“The most interesting AI innovations in healthcare will likely be those that run safely across dozens of sites, wrap around everyday tools like EHRs and RPM platforms, and consistently move hard metrics such as throughput, access, and clinician well-being.”

In Part Two of our four-Part 'AI in Healthcare' coverage, we will investigate the different types of risk as each pertains to major stakeholder groups . . . and we'll look at what the different governments around the world are doing to regulate the requirement for mitigation of these.

Why Invercargill? The Strategic Logic Behind New Zealand's Bid to Become Asia-Pacific's Most Secure AI Data Hub

Thu, 12 Mar 2026 09:35:49 GMT

Why Location Is Everything for New Zealand's $3.5 Billion AI Megaproject

The approval of New Zealand's first AI factory holds genuine international significance - not least because Invercargill, New Zealand's southernmost city and one of the most remote urban centres in the developed world, improbably addresses three of the most pressing infrastructure and location anxieties facing governments and AI projects everywhere.

Datagrid's newly-consented "Data Centre Park" mitigates those anxieties - challenges plaguing governments and enterprise AI strategists alike - in one highly strategic project - a project that mitigates the urgent need for geopolitically secure data infrastructure, the escalating energy cost of AI at scale, and the chronic vulnerability of existing digital backbone networks to seismic events, climate exposure, and deliberate disruption.

The NZD 3.5 billion, 280-megawatt AI factory campus - 49 hectares in Makarewa, just outside Invercargill - will more than double New Zealand's data processing capability and will rank among the largest single infrastructure investments in the country's history.

Geopolitical Case

New Zealand's political stability, geographic remoteness, and absence from the world's primary geopolitical fault lines represent the project's primary positioning strength. In an era when governments are actively reassessing where sensitive data processing infrastructure should be located, a jurisdiction that is simultaneously stable, renewable-powered, and physically remote from conflict zones is a premium asset.

Remoteness, however, cuts both ways. New Zealand's existing international submarine cable connections all land in the North Island - the South Island has none. The Tasman Ring Network, developed by Datagrid in partnership with Chorus, the NZX-listed company that built and operates New Zealand's ultra-fast broadband network, addresses that directly. It will deliver the first international subsea cable connection to the South Island, linking Invercargill to Sydney and Melbourne, with a domestic leg running from Auckland through New Plymouth and Greymouth to Invercargill completing the circuit.

Infrastructure resilience - against seismic events, climate exposure, and deliberate disruption — is now a standard procurement consideration for governments and enterprises making long-term AI commitments. In the New Zealand context, seismic vulnerability is the primary consideration. The Tasman Ring's domestic routing runs down the South Island's west coast, sidestepping the Wellington fault zone entirely.

Energy Case

The global AI infrastructure build-out is running headlong into an energy wall. Data centres powering large language models and GPU clusters are consuming electricity at a rate that is straining grids in the United States, Europe and Asia. Governments procuring AI infrastructure are increasingly required to account for energy source, carbon footprint and long-term power cost, not just processing capability.

Invercargill's average temperature of 9 to 10 degrees Celsius reduces cooling costs substantially. New Zealand's grid runs at 86 percent renewable. Datagrid's campus will sit close to the country's largest hydroelectric stations with direct access to four separate national grid circuits for resilience. At 280 megawatts of consumption, this facility will be New Zealand's second-largest electricity user - and the renewable sourcing of that load matters both environmentally and commercially to the hyperscale AI and cloud providers the campus is designed to serve.

A Team That Has Done This Before

Datagrid's founder and CEO Rémi Galasso previously founded Hawaiki Submarine Cable - the first independent transpacific cable linking Australia, New Zealand and the United States.

Senior Advisor Georges Krebs was the architect of that cable.

Together they have already connected the Pacific once. Now they are building its AI infrastructure layer.

Head of Technology Benjamin Black co-authored the original paper proposing Amazon Web Services cloud architecture.

Head of Power Vince Hawksworth retired as CEO of Mercury NZ in August 2024 after leading major renewable projects including the Turitea Wind Farm.

Head of Sales Alfred Au Yeung, based in California, previously served as Head of Strategic Projects and Network Infrastructure Investments at Telstra, Australia's largest telecommunications company.

This is not a team assembled around a speculative concept.

One Further Detail

As of early 2024, Invercargill was reported as a front-runner for a fibre optic cable connection to Antarctica - a development that, if it proceeds, would add a further dimension to the city's emerging role as a southern hemisphere digital infrastructure hub.

The current status of that proposal warrants watching.

Governance Question

Datagrid is building infrastructure to serve global hyperscale clients - but that infrastructure sits on New Zealand soil and operates under New Zealand law. Every workload processed in Makarewa is subject to New Zealand's privacy framework, data governance rules and AI regulatory environment.

Global clients locating sensitive workloads here are making a bet on the maturity of that regulatory environment as much as on the quality of the physical infrastructure.

Singapore's experience as a global data hub is instructive. Its reputation as a trusted destination for sensitive enterprise and government workloads was built not on connectivity and power alone, but on the parallel development of a governance framework that sophisticated clients could rely on.

New Zealand has the infrastructure ambition. The question is whether its data governance and AI regulatory frameworks are developing at a pace that matches what is being built in Makarewa.

Sophisticated enterprise clients doing due diligence on where to locate sensitive workloads will factor regulatory maturity into that decision. If New Zealand scores poorly on that variable, some will go elsewhere.

That's a dimension Government AI World w ill be watching closely.

Identity Resilience Imperative as AI Wave Floods Workplaces with Agents

Sat, 15 Nov 2025 22:54:48 GMT

Newly-Released Research Shows 90 Percent of Global Leaders Cite Identity Attacks as Top Concern

New research from Rubrik Zero Labs has uncovered a troubling gap between the expanding identity attack surface and organisations’ ability to recover from resulting compromises.

The security and AI operations vendor's report, 'Identity Crisis: Understanding & Building Resilience Against Identity-Driven Threats' , warns that the wave of AI agents washing over government and corporate workplaces equates to a surge of both non-human identities (NHIs) and agentic identities - and is resulting in an urgent need for CIOs and CISOs to identity threats and recovery strategies.

Rubrik's report shows that as AI adoption expands across organisations worldwide, enterprises are taking decisive action to strengthen identity resilience.

Illustrating this trend:

89% of organisations plan to hire professionals within the next 12 months specifically to manage or improve identity management, infrastructure, and security.

87% of IT and security leaders actively plan to change Identity and Access Management (IAM) providers or have already begun the process.

58% cite security concerns as the primary driver to switch IAM providers.

"The rise of identity-driven attacks is changing the face of cyber defence," said Kavitha Mariappan, Rubrik's Chief Transformation Officer. "Managing identities in the era of AI has become a complex endeavour, especially with the labyrinth of NHIs.

"We have an under-the-radar crisis on our hands, whereby a single compromised credential can grant full access to an organisation's most sensitive data. Attackers are no longer breaking in, but logging in.

"Comprehensive identity resilience is absolutely critical to cyber recovery in this new landscape."

Agentic AI Opens the Door to New Identity Challenges

As organisations integrate agents into their workflows, NHIs will continue to outpace the growth of human identities.

Industry reports contend that NHIs now outnumber human users by 80 to 1.

Mariappan contends that "securing NHIs will become as essential as securing human identities - if not more so - given the growing complexity of managing AI agent operations."

Rubrik’s research found that:

89% of respondents have fully or partially incorporated AI agents into their identity infrastructure, and an additional 10% have plans to do so.

Over half of IT security decision makers (58%) estimate that in the next year, 50% or more of the cyberattacks they deal with will be driven by agentic AI.

Weakening Confidence in Recovery Strategies Highlights Need for Identity Resilience

With identities the keys to access an organisation’s most sensitive data, IT and security leaders must build resilient identity services and infrastructures to ensure a quick recovery and restoration of operations in the face of an attack, Mariappan says.

Yet overall confidence in recovery times is declining at a rapid rate, according to the company's research:

In 2025, only 28% of respondents believed they could fully recover from a cyber incident in 12 hours or less, compared to 43% in 2024.

Over half (58%) of respondents believe it would take at least two days to recover and achieve full-service operations, post-compromise.

Of those who experienced a ransomware attack in the past year, 89% paid a ransom to recover their data or stop the attack.

Organisational concern over the state of identity security is valid, and IAM tools alone are not enough to properly address these challenges. CIOs and CISOs need a comprehensive identity resilience strategy for when, not if, an attack strikes.

Research Methodology

The Rubrik Zero Labs Survey was conducted by an external research firm and surveyed 1,625 IT security decision-makers at companies with 500 or more employees. The research was conducted with a 50/50 split between Directors/VPs and CIOs/CISOs.

The research was conducted in three regions: the U.S., EMEA (UK, France, Germany, Italy, Netherlands), and APAC (Japan, Australia, Singapore, India).

Surveys were completed between September 18 and September 29, 2025, using an email invitation and an online survey.

About Rubrik

Rubrik is a security and AI operations company leading at the intersection of data protection, cyber resilience, and enterprise AI acceleration.

Rubrik Security Cloud delivers complete cyber resilience by securing, monitoring, and recovering data, identities, and workloads across clouds.

Rubrik Agent Cloud accelerates trusted AI agent deployments at scale by monitoring and auditing agentic actions, enforcing real-time guardrails, fine-tuning for accuracy, and undoing agentic mistakes.

AI Adoption: Why A Weak Organisational Foundation Guarantees Project Failure

Sat, 08 Nov 2025 06:42:14 GMT

Unreadiness:
The Ultimate Political and Financial Liability

The urge to grab the latest AI fix for chronic government sluggishness is huge. But AI adoption fails a lot.

The problem is rarely the tech itself; it's the agency's lack of procurement maturity - the basic readiness to actually deploy and sustain Artificial Intelligence once the contract is signed.

Buying complex AI solutions when the state of foundational readiness is weak, is a near-guaranteed way to waste taxpayer money. In the currently tight economy of almost every western nation, the public is becoming increasingly sensitive to public funds wastage.

Project failure stands to severely damage public and stakeholder confidence in an agency, or even the government as a whole, if the application and the issue is big enough.

The Four Pillars of AI Maturity

An agency's AI maturity is defined by its readiness across four key pillars.

It's incumbent upon procurement professionals and other relevant members of an evaluation committee to assess these pillars as non-negotiable operational pre-requisite "readiness factors" for any vendor partnership.

1. Strategic Alignment: Solving the Right Problem

Many organisations fall into the trap of purchasing AI for AI's sake. But AI is a tool, not a strategy .

The Solution Trap

Success begins with defining a clear, measurable public service outcome.

Procurement teams must challenge vendors to define the exact public benefit that the AI solution will deliver , as well as to demonstrate a thorough risk analysis and mitigation strategy.

Both the vendor and the agency must ensure the projected expenditure targets a systemic problem, not just a superficial upgrade.

The Operational Mandate

Alignment must combine top-down policy goals with bottom-up operational insights to ensure the solution actually works "on the ground" for the taxpayer.

2. Technical and Human Literacy: Managing the Skills Gap

AI tools don't replace human experts; they demand them.

The Talent Crisis

According to the 2024 Salesforce Public Sector Survey of global government AI professionals, 60 percent report that a lack of relevant skills hampers their ability to apply AI solutions.

Maturity means staff must have the literacy to operate the AI and, even more importantly, the critical thinking capacity to interpret its output and spot flawed decisions.
The Operational Mandate

Agencies must budget for continuous upskilling.

Investment in AI must be immediately matched with a dedicated and fully funded human capital strategy to train staff on the new tools and the data science principles that underpin them.

3. Data Governance: Beyond the One-Time Clean-Up

The fundamental principle is simple: "Garbage In, Garbage Out" (GIGO).

A 2024 global survey of data professionals by Precisely and Drexel University reported that an estimated 70 percent of organisations struggle to trust the data they rely on.

Maturity means setting up continuous data governance policies , not just a single, initial cleaning event.

The Operational Mandate

Agencies must track data lineage (where data comes from and its integrity) and auditability.

Without clean, reliable data, the AI simply automates GIGO faster, making the agency less efficient and compounding the cost of failure.

4. Legacy Infrastructure: The Integration Tax

Legacy systems are the silent killer of AI projects.

The Hidden Cost

The U.S. Government Accountability Office (GAO} has repeatedly highlighted in audits of major Federal departments that up to 80 percent of IT budgets go towards maintaining aging legacy systems.

Every new AI tool must connect to this aging core, incurring a high "integration tax".
The Operational Mandate

Maturity requires a rigorous audit of existing legacy IT systems to predict the integration cost before contracting.

If the new AI solution adds complexity or creates new data silos, the "efficiency" promised by the vendor will be eaten alive by the integration cost.

A Phased, Operational Approach

For vendors chasing high-value government contracts, the pitch must move past simple technological capability.

The modern strategic vendor partners with the agency to diagnose its maturity level before proposing a solution.

Agencies must adopt a phased, operational approach; initial spending should focus on preparing the foundation (data, skills, systems). Only when this foundational readiness is confirmed should the full technological solution be deployed.

This maturity assessment must become a mandatory pre-condition in the Artificial Intelligence procurement process, shifting the focus from buying tech to building sustainable, intelligent operational capacity that tru ly delivers for the taxpayer.

Automating Discrimination: The Unseen Risk in Government Data

Sat, 08 Nov 2025 04:45:06 GMT

Algorithmic Bias:
The Ultimate Procurement & Risk Management Failure

The promise of Artificial Intelligence in government is compelling, but AI merely automates and amplifies existing human processes. Failing to audit for bias of the underlying training data (which directly shapes the AI's logic) and the resulting decision-making model, risks simply (but potentially devastatingly) automating any human-level discrimination . . . at industrial scale.

Algorithmic bias is a clear and present danger that needs to be understood by public sector decision-makers, contract managers, and technology policy teams. For these parties, the danger is amplified because the ultimate, legal liability for discriminatory, non-transparent decisions falls directly onto the public agency, not on the technology itself.

As procurement agencies adopt AI's tools for high-stakes decisions (from resource allocation to predictive policing), the risk of algorithmic bias represents an unseen, massive legal and ethical liability.

But before looking at how it occurs, it's essential to be crystal clear on what it is : Algorithmic bias is the systematic and unfair discrimination inherent in a computer system's results.

It manifests when an AI's system produces results that consistently favour or penalise one group over another, even if the user never explicitly programmed the bias .

How Algorithmic Bias Occurs

Algorithmic bias occurs when the historical prejudices present in the training data (data often reflecting historic societal inequities) are amplified by the AI's system, leading to unequal or unfair outcomes for specific demographic groups.

By way of example, a system designed to allocate resources for social services might learn from data that certain neighbourhoods have historically received a lesser level of government investment, leading the AI to systematically under-allocate resources to those same areas today.

Or, by way of another example, in a hiring tool, if historical hiring data demonstrates a favouring of male applicants, the AI will learn this bias and continue to penalise female applicants, even if all other qualifications are equal.

Profound Risk Management Failure

We're not talking merely about isolated coding errors - which are addressable technical issues.

We're talking about profound, systemic risk management failure compounded by a lack of transparency - a serious management liability.

This is how it works (or perhaps more aptly put, doesn't work):

A lack of transparency means the process is a "black box".

If an agency uses a biased AI system to deny a housing application, and the agency cannot explain why the system reached that decision, the agency has failed its legal and moral duty.

While, on the one hand, this turns the AI into a legal liability shield - a way for the agency to claim, "The computer did it" , and skirt accountability for a discriminatory outcome - a responsible agency and any forward-thinking senior management personnel will clearly see the increasing magnification of what would now be baked in as a systemic issue for that organisation.

Accountability Should Be Integrated into the Procurement Process

The risk management failure stems from neglecting to mandate accountability and auditability, which - critically - should be integrated directly into the procurement process.

This means that the required technical controls - the ability to trace a decision back to its source and verify its fairness - must be written into the Request for Proposal (RFP) and the final contract.

By signing a contract without these provisions, the agency legally ties its hands, making it impossible to enforce ethics after deployment.

The Implications for Procurement Policies & Processes

The moral of the story for procurement leaders and vendors is that the focus must shift from simply evaluating speed, to rigorously auditing the AI’s social and ethical impact. And that starts with explicit, mandatory requirements for auditability and accountability written directly into the RFP.

This must sit as a pre-requisite for deployment, assessed before technical speed or cost.

The ramifications of integrating Artificial Intelligence trained on undetected disparate, biased data, into agency decision-making processes, are immediate and extreme: from loss of public trust all the way through to legal challenge under anti-discrimination laws, and political backlash.

Radical, Auditable Transparency: Nothing Less Will Do

Clearly, it's about prioritising prevention as a strategy, over cure.

At agency level, there must be auditable transparency of all data sources.

At the vendor's end, they need to "open the black box" and allow the agency client access to the model's decision-making architecture or "reasoning" i.e. the ability to track why the model made a decision (explainability) and to verify the data that was used to teach the system.

Without an explicit review of the training data and the detection of any and all potential for unfair weighting of different factors in the decision-making process, any government AI initiative runs the risk of violating anti-discrimination laws and eroding public trust.

Securing full contractual accountability requires mandatory, verifiable standards to become non-negotiable for all contracts and policy implementation.

Who Bears the Burden of Obligation? The Vendor’s Mandate

It's a common error for public sector agencies to assume that the responsibility for ethical performance resides solely within their own compliance team.

In the case of AI accountability, this assumption is flawed. While the agency bears the ultimate public liability for the system's discriminatory outcomes, the burden of obligation - the capacity to demonstrate ethical compliance - must be contractually held by the vendor.

The vendor is the only party that owns the intellectual property (IP) for the AI model, which includes access to the proprietary training data, algorithmic weights, and the decision logic - the core elements of the "black box".

The government buys the right to use the software, but the vendor retains control over its internal mechanics .

Therefore, for the procurement contract to be legally sound, the agency must make the vendor contractually liable for proving the system's ethical integrity. Without this specific clause, the government assumes the entire legal risk for a system it cannot inspect.

Mandatory Contractual Mechanisms

To manage this risk, the RFP and the final contract must contain clear and specific provisions that mandate that the vendor provides the necessary tools and documentation for compliance.

The rationale for placing this burden on the vendor is simple: the vendor possesses the technical superiority to manage the risk .

While the agency owns the historically flawed data, the vendor is the expert in machine learning, and they alone can detect, measure, and build mitigation strategies for biases.

The government's contract must stipulate that the vendor is responsible for these key actions:

Audit Tools and Explainability: The vendor must deliver specific mechanisms that allow the agency's auditors to track the AI's decision path (explainability). This is the technical capacity to trace a decision back to its source data.

Bias Reporting and Data Provenance: The vendor must warrant that the training data has been subjected to a rigorous bias assessment and must provide the results. Furthermore, they must fully document and trace the source of all training data to ensure it meets public sector standards for completeness and impartiality.

By mandating these provisions, the government shifts the vendor's role from simply selling technology to an agency, to guaranteeing ethical accountability. If the vendor cannot provide these contractual guarantees, the agency must not sign the contract.

This final check is the only way to ensure the public's trust remains intact and that Artificial Intelligence serves as a tool for genuine progress, not automated injustice.

The Bespoke Citizen: Estonia's Masterful Evolution of AI-Driven Governance

Sun, 19 Oct 2025 07:23:29 GMT

Estonia . . . Where IS That? And Who Would Have Thought?

When you think of global leaders in artificial intelligence, you might think of the United States or China. But when it comes to the practical, nation-wide integration of AI into government services, the undisputed frontrunner is a small Baltic nation most of us know little about: Estonia.

Often overlooked on the world stage, Estonia has quietly become the digital gem of Northern Europe – a highly developed, high-income country where technological innovation isn't just a policy, but a national identity.

The country's tiny 1.3 million population, spread across a small, heavily-forested landmass, goes a long way towards explaining why a digital-first communication strategy was so appealing to its leaders.

A member of the European Union, NATO, and the OECD, this little geographic jewel was under occupation by the Soviet Union until 1991. When it regained its independence, it essentially had to build a modern government from scratch.

The bright side, though, was that it had no legacy infrastructure to hold it back.

That clean slate was its greatest advantage. With no archaic paper-based systems and no entrenched bureaucracy, Estonia's leaders of the mid-1990s made a radical strategic decision: they committed to building a lean, digital-first state rather than a large, expensive physical bureaucracy.

e-Estonia . . . A (Very) Early Starter

For Estonia, this was never just a technology project; it was a nation-building philosophy. For the ensuing three decades, successive leaders have systematically built the world's most advanced digital society, often known as e-Estonia .

But is this unique to them? While many countries have excellent "digital front doors"—Singapore's world-class digital identity program, for example - none have achieved Estonia's level of deep, legally mandated data integration.

The key difference lies in two foundational pillars: the "once-only" principle , which forbids the government from asking a citizen for information it already has, and the "X-Road" , the secure technology that allows different departments to exchange that data seamlessly. Because of this, most analysts conclude that Estonia remains years ahead of the international pack.

And Now . . . Introducing 'Burokratt'

The latest evolution of this philosophy is the AI-driven virtual assistant network, "Bürokratt" .

Officially launched into its development phase in 2021, its goal, according to the Estonian Ministry of Economic Affairs and Communications, is to provide refined, seamless, and proactive public services.

Reflecting the adaptive nature of AI ecosystems, the Bürokratt initiative was not a single-event launch. Following its initial development phases, the project has now, in 2025, entered an advanced pilot stage, testing specific "life event" scenarios with the goal of a gradual roll-out over the coming years.

The catalyst for Bürokratt was the growing incongruity between Estonia's already-advanced digital governance environment, and the remaining friction points in citizen services. The goal was to eliminate this final friction point by creating a single, pro-active interface.

According to Ott Velsberg, the Chief Data Officer of the Government of Estonia, the vision is clear:

"The ultimate aim of Bürokratt is that a citizen will no longer have to wander between different institutions. The state will have a single virtual face."

The state, says Velsberg, “should know what is happening in the citizen’s life, and the state should be offering the services to the citizen, not the other way around."

Trust, Security & the 'Glass Box'

It’s statements like that, however, that land awkwardly with privacy advocates – and, indeed, with most citizens in Western democracies. And it raises the single most important question: why do Estonians allow this?

The answer is the secret sauce to their success: they built public trust by engineering a system of radical transparency, a "Glass Box" model , not a "Black Box." The system is built on the principle that the citizen owns their data , and the government is merely a custodian.

In practice, every citizen can use a tool called the " Andmejälgija" (data tracker) to see a clear, un-editable log showing exactly which official has accessed their data, and for what reason, in real-time. This transparency is backed by severe legal penalties. Under the Estonian Penal Code, unlawful access to data is a criminal offence punishable by hefty fines or up to five years' imprisonment.

Thus, trust is not just promised; it is engineered into the system, by giving citizens total visibility and control . . . and, in so doing, ensuring peace-of-mind uptake and usage by the vast majority of the citizenry.

The system aims to transform government from a frustrating bureaucracy into a helpful life-admin partner. A citizen turning 65 could be proactively asked, "It looks like you're eligible for your state pension. Would you like us to start the application?" It also promotes equity by closing the "knowledge gap," ensuring everyone gets the benefits they are entitled to, not just those who are savvy enough to ask.

However, understandably, this level of data integration also raises profound questions about the "Surveillance State." A government that has a complete, real-time picture of a citizen's entire life is, arguably, all-powerful. Even in Estonia, the questioning attitudes will almost certainly increase over time, as the international debate heats up regarding the line between a helpful, proactive government and an overly intrusive one.

The trade-off between pro-active convenience and potential surveillance is the central question that will define the future relationship between the citizen and the state in the AI era. It's a complex and controversial frontier that Government AI World will be exploring in-depth in future coverage.

The AI Procurement Paradox: Why the Safest Playbook is Now the Riskiest

Sat, 18 Oct 2025 03:52:05 GMT

Government's traditional playbook was designed to buy known quantities . . . but it's going to fail for innovative AI procurement.

Governments and public sector organisations worldwide are confronting the same systemic challenge: how to strategically procure Artificial Intelligence. The traditional playbooks are failing, and the search is on for a new model.

Nowhere has this challenge been brought into sharper focus than in the United States, where a recent White House mandate provides a powerful case study in the shift from diagnosis to direct action.

In August 2025, the United States’ White House Office of Management and Budget (OMB) issued a landmark memorandum, ‘M-25-15: Advancing Governance, Innovation, and Risk Man agement for Federal Agency Use of Artificial Intelligence.’

In line with the general speed of initiative roll-outs in all categories across the Trump Administration, ‘M-25-15’ isn't intended to be the quintessential dust-gathering report; it's a direct order. It mandates that by December 1, 2025, every Federal agency must implement specific safety measures and governance structures for their AI systems.

An even more “hot off the press” report – the September 2025 U.S. Government Accountability Office’s (GAO) "Artificial Intelligence: Agencies Report Progress, but Need to Fully Implement Key Accountability Practices" - amps up the pressure further.

In short, the OMB memo has set the policy and issued the deadline, but the GAO document has highlighted the arguably chasmic gap between what agencies and departments are being ordered to do, and what they actually have the capability and the smarts to do.

Central to bridging that chasm is genuinely strategic procurement: acquiring the right AI capabilities in the first place. Not just some fuzzy use of the word “strategic.” Actually strategic. And strategic, in the context of Artificial Intelligence – with its characteristically uncertain outcomes, continuously evolving capabilities, and deep dependence on an agency's own unique data – does not necessarily carry the same definition, thinking, and set of processes as the procurement of well-defined goods and services with fixed specifications.

Deterministic vs Probabilistic . . . Procurement Professionals Have Been Struggling

Procurement professionals – and, ironically, even seasoned IT procurement pros – have been struggling . . . struggling to determine what they need, to articulate what they need, resultantly to actually procure what they need, and then ultimately to ensure what they bought, keeps meeting the required outcomes going forward. Their (certainly, hard-won) expertise is in de-risking the purchase of complex but ultimately predictable systems, defining a known quantity and required outcome specifics down to the last micro-detail. But AI is not a known quantity. It is not a static piece of software with a finite feature list; it is an evolving capability that learns from, and creates its outputs from, data.

To re-frame the same problem: Traditional software, no matter how complicated, is deterministic . It operates on a fixed set of pre-programmed rules. In the most general sense, given the same inputs, it will always produce the same output. The very essence of AI is probabilistic . It doesn't follow rigid rules; it identifies patterns in data to generate its outputs. Thus, the expertise required to buy a predictable, rule-based system is fundamentally different from the expertise needed to acquire an evolving, pattern-based capability .

In short, this new domain comes with highly unfamiliar core technical realities – and navigating the strategic implications of those realities will be the root and the starting point of successful AI procurement outcomes.

The Architecture of A New Playbook

Diving deeply into the problem and spending enough time evaluating it begins to suggest exactly what the strategic navigation of the component challenges would look like.

IIn its 2022 report, ' Why government struggles to buy AI: And what to do about it' , influential think tank, the Brookings Institution, provided a deep-dive, clear diagnosis of the problem in no uncertain terms, essentially demonstrating that the government's traditional playbook for buying technology is fundamentally broken when it comes to AI.

These traditional methods are not prescriptive enough to guarantee safety, yet not flexible enough to foster true innovation. And certainly, overall, not agile enough . Not agile enough to manage uncertainty, to adapt to evolving technology, or to engage with the innovative start-ups that are building the future.

The Brookings report made clear that this is a systemic challenge that’s been quietly building for years. The evidence for this long-standing, structural mismatch lies in their core finding: "Traditional procurement methods, which are designed to acquire well-defined goods and services, are ill-suited for AI, which is an evolving capability that requires experimentation and iteration."

The New Playbook: From Procuring 'Things' to Procuring Outcomes

Speaking as a strategist, once the problem is defined this clearly, the architecture of the solution starts to reveal itself. And it's clear that consistent, sustained success is going to require a fundamental paradigm shift.

The traditional RFP asks for answers . An AI-ready procurement process must learn to ask brilliant questions . It moves from, "Here is the exact thing we want; give us a price" to "Here is our core operational challenge or strategic opportunity; propose how your AI capability can solve it."

Procurement professionals can, though, take comfort in the fact that this isn't about abandoning certainty. It’s about building a new kind of certainty – a certainty based on process and performance , rather than on fixed specifications.

Building 'New Certainty'

So what would that process look like in a procurement, in practical, in-action terms?

A search for, and investigation of, the ideal solution would, arguably but very logically, start here: Instead of a single, massive contract based on promises, modern procurement builds confidence through three key phases:

Certainty of Viability:

Before committing to a multi-million dollar program, agencies can award multiple smaller, fixed-price contracts for 90-day pilot projects. The goal isn't a finished product; it's to answer one question: "Can this vendor's approach actually work with our messy, real-world data?"

The certainty being purchased is proof. This "try before you buy" approach, championed by government innovation hubs like 18F, a technology and design consultancy within the U.S. General Services Administration (GSA), de-risks investment by making "small bets" to find out exactly what works – and what doesn’t – before scaling.

Certainty of Performance:

With a viable partner identified, the contract would logically shift to focus on outcomes – but the what , or the satisfactory nature of the results, rather than the how . As opposed to the conventional "build a system with X features" approach, it's "deliver a system that achieves Y result" contract.

For example: "The final system must identify fraudulent transactions with 95 percent accuracy on live data." The certainty being bought is a performance stipulation . Thus, it’s still a non-negotiable measurable .

Certainty of Cost:

No procurement professional wants to sign a blank cheque. Structuring the overall contract in delivery-and-projectable evolution phases, each with a clear budget control, would provide the guardrails against doing so.

For example: "Phase 1: Deploy to the first department for $X. If performance metrics are met, we will then authorise funding for Phase 2."

The certainty being purchased is budgetary control , ensuring the next step is funded only if the previous one was a proven success.

Managing An Evolving System: Guardrail Specifics

The continually, and rapidly real-time, evolving nature of Artificial Intelligence makes it clear no contract award will end up being a one-shot, “implement-and-call-us-if-you-need-us” one-time transaction (unless your procurement and your provider were indeed a lemon); it's going to be, quite essentially, the beginning of a structured partnership.

The procured “solution” (to resort to standard IT terms that don’t quite fit the new AI dynamic) is a "living" capability that will continue to evolve. To ensure it evolves in the right direction, a framework of human-centric oversight is essential.

Logically, this would include at least the following policies and practices:

Constant Performance Monitoring:

The contract must specify that the vendor provides a real-time dashboard tracking the AI's vital signs (e.g., accuracy, false positive rates). If a key metric dips below the agreed-upon threshold, it triggers an alert, obligating the vendor to investigate and fix it before a major failure occurs.

Human Governance:

The agency or department must establish a cross-functional governance team – an 'AI Board', if you like – to provide strategic and ethical oversight. This board would meet regularly to ask the “big questions”: Is the AI's advice still aligned with our agency's policies? Has it developed any unforeseen biases?

Safe Testing:

The golden rule of AI procurement appears to be never to let a more advanced or updated version of the AI be deployed into the live operational environment without testing it first.

For example, when a vendor proposes a new, "smarter" version of the AI, that new model must first prove it is better than the current one. This is typically done in a safe, offline "sandbox" environment, where the new version is run against recent historical data and its results are compared directly to the old version's. The update is only deployed if it is demonstrably better. It's the equivalent of a flight simulator for your AI.

In a forthcoming article, I’ll write about the powerful real-world example of this new approach as provided by the United States Defense Innovation Unit (DIU) and its "Other Transaction (OT) authority", which has compressed the procurement cycle to see it award prototype contracts in a stunning 60 to 90 days – proving that speed and governance can co-exist, an essential in light of the new U.S. governmental mandates.

‘Smart’ Traffic Management: A Tale of Two Cities

Sun, 12 Oct 2025 04:33:45 GMT

AI-Facilitated Traffic Flow Proficiency:
One Common Objective. Two Very Different Methods.

It’s one of the most common and maddening moments of modern life for city dwellers everywhere: sitting in a car, staring at a red light, while absolutely no traffic crosses the empty intersection.

For over a century, the traffic signal has been a “dumb” and unthinking tyrant, dictating the flow of our cities with simple timers and rudimentary sensors. It’s cost time and fuel, and tested patience.

But progressively-minded cities are increasingly relegating this rigid, one-size-fits-all approach to the pages of analogue history. Forward-thinking traffic management engineers and executives have been asking: What if that light wasn't so dumb? What if it could see the approaching traffic, communicate with the next intersection, and make intelligent decisions in real time?

And some have been leading the way in making that a reality, deploying Artificial Intelligence to transform their traffic control systems from mindless, timer-based apparatuses into responsive, thinking networks.

Two case-studied cities and their two very different approaches within this overarching new paradigm are the subject of this deep-dive: Singapore, and Pittsburgh in the United States.

Their goal has been the same in each case: untangle gridlock and create smarter, more efficient urban arteries. Their methodology and its implementation, however, has not.

This is a tale of two fundamentally different philosophies not just of technology, but of strategic execution: Pittsburgh’s model of rapid, university-born innovation versus Singapore’s masterclass in long-term, state-driven evolution.

Singapore ‘GLIDES’ Into ‘Top-Down’ AI-Facilitated Future for Traffic Management

Singapore’s Land Transport Authority (LTA) has opted to deploy Artificial Intelligence-assisted traffic management according to what might be described as a “top-down” approach.

Key in the masterminding and roll-out of its new AI-driven “GLIDE” (Green Link Determining System) has been a priority focus on system-wide benefits, as opposed to percentage point improvements in any single metric. The system is designed to improve the efficiency and reliability of the entire road network .

It's about making journey times more predictable for everyone by optimising efficiencies in a holistic sense.

GLIDE’s logic prioritises, for instance, the smooth, predictable flow of the overall system over achieving the absolute fastest travel time on one specific road if that speed comes at the cost of wider network disruption.

To this end, the centralised command-and-control nature of GLIDE is foundational to its role and function within the much larger, nation-wide "Intelligent Transport Systems" (ITS) framework of this densely-populated, operationally efficient city-state, renowned for its long-term planning.

While the foundations of Singapore's system aren't freshly-minted, their relevance today is more potent than ever. In the age of AI, its two-decade head start has created an unmatched strategic asset: a vast trove of historical traffic data. This "data moat" is the fuel for today's sophisticated predictive AI, allowing the city-state to model and manage its network with a maturity other cities can only study. The story of GLIDE, therefore, is less about a single piece of technology and more about the powerful lesson it offers in long-term governmental vision and data governance.

A Strategy of Total Information Awareness

The operational concept behind GLIDE is to achieve network equilibrium .

In this centralised model, a single core processing system at the LTA's Intelligent Transport Systems Centre creates a complete, real-time digital picture of the road network, allowing it to manage traffic proactively rather than reactively. Analogically-speaking, the system acts as a single, all-seeing air traffic controller for the city's entire road network.

The methodology is one of total information awareness. Data from thousands of sensors embedded in the road surface, which detect real-time traffic volume and speed, is fed continuously into this central command hub.

GLIDE’s central computational engine analyses a network-wide traffic model, conducting real-time traffic modelling and predictive analyses, to optimise the phasing and timing of traffic signals for the entire network. It then makes strategic adjustments to signal timings to ensure the smoothest possible flow across the entire island.

The goal isn't simply to clear one congested intersection; it's to create the most efficient state for the system as a whole.

Integrated Public Service Capabilities

One of the most significant advantages of this centralised model is its ability to integrate with other public services.

The “holistic” philosophy extends to actively supporting public transport, where giving a bus slight priority can improve the reliability of the entire service, encouraging its use and ultimately reducing total vehicle volume.

GLIDE achieves this through its EMTRAC (Emergency Vehicle and Bus Priority) component system, which can also create uninterrupted passage for emergency vehicles responding to a call.

This top-down control also allows traffic managers to dynamically respond to major incidents, like a highway accident, by proactively re-routing traffic and adjusting signal patterns city-wide to prevent gridlock. It is a strategy of total command-and-control, designed for holistic optimisation and deep integration with a city's operational needs.

Pittsburgh's 'Bottom-Up' Approach

If Singapore's approach is one of a single, central digital commander, Pittsburgh has championed an almost opposite philosophy: a "bottom-up" revolution in traffic control, where the system acts less like a single commander and more like a co-operative team.

The city’s “Surtrac” (Scalable Urban Traffic Control) system, born out of Pittsburgh’s own world-renowned Carnegie Mellon University, shuns the “centralised, city-wide brain” concept in favour of making each individual intersection “smart”. It's a strategy built on the principles of localised intelligence and adaptive co-operation, whereby collective intelligence emerges from simple, local interactions.

Each Surtrac-equipped intersection acts as its own autonomous decision-maker. Using real-time data from its own radar and video feeds, the local AI computer analyses the approaching traffic flow.

Second-by-Second Predictive Modeling for Optimal Signal Sequence

The ingenious component is how it then develops the optimal plan: using a second-by-second predictive model, it chooses the signal sequence that will result in the minimum aggregate delay for all vehicles it can see. Crucially, it then communicates this plan to its immediate neighbours. I t effectively tells the next light down the road, "I'm sending a platoon of cars your way in 15 seconds," creating seamless "green waves" without direction from any central authority.

The results of this localised, adaptive strategy have been significant. The pioneering pilot projects, with results first published in 2012, recorded a more than 25 percent reduction in travel time, a decrease in idling time at lights by an average of 40 percent, and a corresponding 21 percent decrease in vehicle emissions.

Indeed, these projects provided such a successful proof-of-concept in demonstrating that a de-centralised approach could offer a nimble and highly scalable alternative to a massive, top-down overhaul, that the Surtrac technology has since been commercialised and is being deployed in other cities across North America.

Beyond Security: Why 'Data Use Governance' is the Essential Architecture for Government AI

Sat, 11 Oct 2025 06:47:26 GMT

From Passive Security to Active Governance of Data Usage

Governments worldwide – from the European Union with its AI Act to the United States with its 'Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence' – are racing to establish frameworks for the (genuinely) safe and trustworthy deployment of Artificial Intelligence.

But beneath these high-level directives lies a universal reality that bridges technology issues and management practices: you can't have responsible AI without first mastering the appropriate and agreed ethics and policies related to the collection and use of the data that feeds it.

And that, in turn, requires a new discipline and a new mindset beyond simple data security. It requires smarts and professionalism in data use governance – in direct accordance with watertight ethics, practices and policies.

Government agencies have spent years focusing on data security and related governance . . . protecting citizens’ data from breaches. And yet, compared to the concept-in-action of data usage governance, the “data security” mindset is, to a large degree, passive; “set and forget”.

An Active Discipline

Data usage governance is an “active” discipline. It requires the creation of comprehensive data inventories, the defining of clear rules, and the automated enforcement of those rules.

Only through a comprehensively structured, dynamically monitored, action-oriented continuous improvement approach can public sector leaders fully unlock the benefits of AI – that is, without eroding the trust and confidence that is foundational to any government entity’s relationship with its publics.

The sweeping AI Executive Order from the U.S. White House in October 2023 , under then-President Joe Biden, notably, created a baseline blueprint for an AI Bill of Rights, and put that country’s government agencies under a clear mandate to ensure the AI they build and deploy is safe, secure, and trustworthy.

Traditionally, data governance focused on securing highly organised, structured database-and-spreadsheet type data inputs – stored in a central warehouse for predictable reports. By stark comparison, the new reality of data use governance must control how Artificial Intelligence systems and automated agents access a chaotic mix of data from all corners of an agency – from, for example, structured and unstructured files like emails, PDFs, and reports.

The core problem is that AI models can indiscriminately access and "learn" from any data they can reach, including sensitive and personally identifiable information (PII). Crucially, once learned by many current AI models, this data cannot be easily removed.

It's Not A Tech Problem, It's A Trust Problem

While the private sector worries about brand damage, the stakes for government are even higher: the erosion of public trust.

When an agency's AI systems misuse citizen data, it isn't just a compliance failure; it's a violation of the social contract.

To demonstrate and uphold that social contract of public trust, agencies must govern data based on four key contexts:

Data Context (The "What"): What is this data? Is it a public record or a sensitive health document?

Consent Context (The "Permission"): Did the citizen provide this data specifically for this purpose? Data for a water bill should not be used for law enforcement profiling.

Regulatory Context (The "Rules"): What specific laws govern the use of this data, such as HIPAA or GDPR?

Business Context (The "Why"): What is the specific, legitimate public purpose for this AI to use this data?

Answering these four questions is the only way to enforce responsible use policies. Without them, an agency cannot prove its actions are legal, compliant, or ethical.

And the risk is a loss of the citizenry’s trust that is nearly impossible to regain.

From Data Gatekeepers to Trustworthy, Purpose-Driven Innovators

In short, the transition to AI requires government agencies to shift from being reactive data gatekeepers to proactive enablers of responsible, purpose-driven, policy-centric, privacy-protecting innovation. And a growing consensus is emerging among technology leaders that the old, data-centric model of governance is no longer sufficient to underpin that imperative.

Ultimately, mastering data use governance is not just a technical or compliance challenge; it is a prerequisite for any government wanting to innovate responsibly. As agencies move forward, the choice is not if they will adopt AI, but how they will build the public trust necessary for it to succeed. That foundation is built on a proactive, policy-driven, and purpose-based approach to the data they hold.

Note from Author & Editor-in-Chief, Jordan Kelly:

This is a vast area of intensifying government, vendor and public interest. There’s a lot to unpack. I’ll be doing that in future deep dives and subject matter expert interviews.

New Zealand Govt Launches First 'AI Strategy' - Designed to 'Boost Productivity' & 'Grow Economy'

Thu, 18 Sep 2025 05:33:26 GMT

Science, Innovation & Technology Minister:

'New Zealand must develop stronger AI capabilities to drive economic growth.'

The New Zealand Government Minister of Science, Innovation and Technology, Dr Shane Reti, has launched New Zealand’s first AI Strategy to boost productivity and grow a competitive economy.

“AI could add $76 billion to our GDP by 2038, but we’re falling behind other small, advanced economies on AI-readiness and many businesses are still not planning for the technology,” Reti says.

“We must develop stronger AI capabilities to drive economic growth, and this strategy sends a strong signal that New Zealand supports the uptake of AI.

“The Government’s role in AI is to reduce barriers to adoption, provide clear regulatory guidance, and promote responsible AI adoption.

“We’re taking a light-touch approach. The strategy's function is to set out a commitment to create an enabling regulatory environment that gives businesses confidence to invest in the technology.

New Products & Services, Improved Efficiency, Better Decision-Making

“Private sector AI adoption and innovation will boost productivity by unlocking new products and services, increasing efficiency, and supporting better decision-making.

“New Zealand’s strength lies in being smart adopters. From AI-powered precision farming techniques to diagnostic technology in healthcare, Kiwi businesses can tailor AI to solve our unique challenges and deliver world-leading solutions.”

Reti claims the AI strategy aligns with OECD AI principles and "the Government will continue to work with international partners on global rules to support the responsible use and development of AI".

(Somewhat presumptuously speaking for the country's citizenry) Reti says, "New Zealanders will need to develop trust and give social licence to AI use."

To that end, he says, the New Zealand Government has also released a ' Responsible AI Guidance' publication.

"The Government will use existing legislation and regulations such as privacy, consumer protection and human rights, to manage risk and privacy concerns."

Editor's Opinion: The jury should remain firmly out on whether New Zealand's weak and often seemingly optional consumer protection, human rights, and privacy regulations will provide any protection at all against the misuse or questionable use of Artificial Intelligence by New Zealand government and commercial entities.

Introduction Into Schools Must Be 'Thoughtful Yet Open-Minded' & Highly Strategic

Wed, 03 Sep 2025 23:15:03 GMT

A Massive Upside IF It's Done Right . . . and An Unrecoverable Downside If It's Not

(By Jamie Munro, AI & Robotics Expert)

As it becomes increasingly clear that Artificial Intelligence (AI) will play a major role in the future, the issue of AI in education is beginning to gain the attention of governments around the world.

Early last month, China announced a mandatory eight hours of AI education every academic year for all students aged six and above. A few weeks ago, U.S. President Donald Trump signed an Executive Order instructing schools to begin AI training for staff and students. And just two weeks ago, the UAE announced plans to introduce AI to their national curriculum for all age groups.

Social & Economic Divide Will Be Huge

Governments and school authorities who fail to respond, or even worse. move to ban schools from using AI, risk getting left behind. The social and economic divide between those who master AI and those who don't, will be huge. And countries who fail to prepare their next generation workforces for the technologies of the future could find themselves falling behind.

AI is an incredibly fast-moving technology, and if governments don’t act now to implement some form of early education, not only will their young people struggle to find fulfilling work, but we risk repeating the mistakes of the social media era - in which parents and governments failed to grasp the massive impact of these technologies until it was too late.

That mistake has resulted in major safeguarding and cyber bullying issues, young people failing to develop the social skills expected of them, and a loneliness epidemic among the most “connected” generation in history, who, ironically often struggle to form meaningful relationships.

How much worse could a similar mistake in the AI era be?

Cheating & Over-Reliance: It’s Going to Be A Balancing Act

Equally huge, however, are the issues schools face in adapting to the new reality: "Cheating" using AI tools is rife and teachers must find new ways to assess students.

Students must learn to use these new tools, and yet still develop marketable skills without becoming over-reliant on AI. It’s going to be a balancing act, as the next-generation workforce will need to add value in a world increasingly run on these technologies.

(Add to this picture, the myriad of new safeguarding challenges AI introduces: Biased information, harmful content, fake news . . . these are just the beginning. There have been cases of teenagers falling in love with AI chatbots and other chatbots have even encouraged their users to commit suicide.)

The Era of AI Requires New Tools for Schools

So new tools for schools will need to be developed. And fast.

Schools and governments must resist the urge to simply ban these new technologies. Doing so will not prevent students from accessing them and will increase the likelihood that they will use them in unhelpful ways that actually damage their learning.

In this new era, students will need access to the latest cutting-edge AI technology. They will need AI assistants that gently guide them towards solutions whilst still giving them the room to learn and make mistakes. But it needs to be executed within robust safety mechanisms to prevent inappropriate usage and outcomes, and to enable teachers to effectively monitor student activity.

Some argue that introducing AI into education is a slippery slope, and certainly I’ve only outlined a few of the macro-level concerns. But these concerns must be carefully weighed against the undeniable reality that AI literacy will become another invaluable skill that students must master.

A Massive Upside, IF It’s Done Right

And as far as teachers go, teaching is about more than simply imparting knowledge.

There’s an important relationship between student and teacher that cannot be replaced by machines. The greatest opportunity from the teacher’s own perspective is to use AI to reduce unproductive burdens and make them more effective at the high-value components of their jobs.

There’s a massive upside for students, teachers, schools and governments who embrace the challenge; but it will require the employment of this new technology in a manner that is both thoughtful and open-minded at the same time, and wholly strategic.

________________

About the Contributor:

Jamie Munro is an AI and robotics expert and entrepreneur focused on developing practical, real-world applications of machine learning. He is the founder of the AI-in-education initiative, WillowLearn, and the creative force behind the AnyModel.co platform.

Policy & People

Executives Driving Efficiency & Governments Predicting Voting Results with AI . . . But Where Does All the Data Come From?

Wed, 03 Sep 2025 23:15:01 GMT

AI Multi-Model Expert Delves Into the Ethics Concerning the Sourcing of Data that 'Feeds' LLMs

The last few years have seen Artificial Intelligence break out increasingly into mainstream usage. Its usefulness isn’t the question. The questions, at the user level, revolve around how to employ it for greatest effect (including competitively), and at the level of AI model training, how to advance AI models that ensure balanced, properly informed outputs from ethically-sourced inputs.

Executives are looking to drive efficiency improvements. Democratic governments are starting to use AI to predict voter outcomes while their counterparts in authoritarian regimes are using it to monitor citizen activities and crack down on dissent. Three in five doctors in the U.S. now report using Artificial Intelligence as part of their practice. Internal research at my firm (willowlearn.com) indicates that between 40 and 60 percent of teachers in the UK now use AI in some component of their role. The most eager early adopters of AI are university students - with the latest numbers from the UK's Higher Education Policy Institute showing 88 percent of university students admitting to the use of AI in their assignments.

You’re An AI User Whether You Know It or Not

You've almost certainly started to hear the names of various AI products coming up in conversations - names like ChatGPT, Gemini, Claude and Perplexity.

Even if you haven't deliberately used one of these products, you've definitely interacted with some sort of product or service employing AI in some way. If you've used Google over the last few months, you will have started noticing the "AI overview" section at the top of the results (congratulations – you are now an AI user).

So now that you are an AI user, you might be asking, what exactly is AI? To give a simple answer, “AIs” – in layman’s terms – are computer programs (“Large Language Models” – “LLMs”) that you can interact with using natural human language. Much like texting with a (very knowledgeable) friend. When you send a message to an LLM, it will respond back to you with its own message. What it's really doing is using some very complicated mathematics to predict, based on your input, what it is that you want to see, and then generating it for you. It's a bit of a simplification, but you can think of it as a beefed-up version of the predictive text system on your phone.

So How Are These Computer Programs Able to Predict What You Want to See?

This is where data comes in, and lots of it. AI models need to be shown millions (and trillions) of examples of human language in a process known as "training". This leads to another question - where does the data come from?

The answer to that question is sensitive.

Firstly, companies developing Artificial Intelligence LLMs for the marketplace need vast amounts of data to remain competitive. Many have been accused of employing unscrupulous methods to obtain it.

The issue of copyright around AI training materials is still an open question and probably one that can only be addressed with new legislation. It emerged earlier this year that Meta (Facebook’s parent company) used a large number of pirated books as part of its training data. That's just one of many copyright cases against AI giants currently making their way through the U.S. Courts system.

Copyright Dilemmas: Is It Fair to Content Creators & Other Humans?

Proponents of greater copyright protections would argue that AI fundamentally breaks the current monetisation systems for content creators.

Under current systems, content creators list their content on search engines like Google and YouTube. Users search for topics they are interested in, the search engine shows the user adverts (which is how they make money), then the user clicks on an item that interests them. The creator of that content then has the chance to monetise their readers/viewers.

As users move away from search engines and towards Artificial Intelligence, AI services simply provide an authoritative answer to the user’s query without linking to an external resource. Previously, a user searching for “how to change a tyre” might find their answer on a mechanic’s website and decide to purchase some service from that mechanic’s business. But in the future, they will increasingly ask an AI how to change the tyre, and the AI will just tell them - even if the AI originally learned how to do that from the mechanic's website.

Is that fair to the original content creator (the mechanic, in this case)?

Proponents of reduced copyright protections would argue that the above argument would never be applied to humans. if I spent a year reading books about a particular topic and became an expert, and then I wrote my own book about the topic, nobody would argue that I'm just regurgitating the books I read (unless I was found to have plagiarised large chunks without attribution).

The other argument is economic: if we start enforcing copyright protections for the creators of AI training material, the cost of these already expensive AI models will get even higher – and countries that don't care about copyright protections (such as China) will gain a massive advantage.

Concentration of Control & Potential Misuse of Power

The second major data-related problem is one of control and power.

Everything an AI "knows" comes from its training data, so whoever decides what is included in the training data has immense power.

In a world where everyone gets their information from AI, the creators of AI will have massive influence over public opinion, much like the broadcasters and newspaper owners in the pre-social media world. Currently this power is concentrated in the hands of a few tech giants like Google, Meta and OpenAI.

Efforts to regulate these giants and AI more broadly could result in concentrating the power into even fewer hands.

AI Bias in Elections? And Why Do They All Sound the Same?

There have already been numerous accusations of bias made against major AI models; during the 2024 U.S. elections, for example.

Many users note that the major AI models, even ones from different developers, end up producing rather similar "facts". AI models are only as good as the training data and any biases in this data will certainly be reflected by the model. The similarities between model outputs can be explained if you consider that all the AI companies are broadly using the same training data – the contents of the internet.

The training data supplied to an AI model acts as a type of voting mechanism. The more times a particular idea or concept appears in the training data, the more likely the model is to reproduce it. This means that AI models are much more likely to adopt mainstream viewpoints, and less popular, more controversial or dissenting opinions and views are much less likely to be represented.

Are we heading towards a future where ordinary people only have access to the “official line”? And do we want the likes of Google, Meta and OpenAI deciding what that official line should be?

It’s too late to go back to a world without AI – that ship has already sailed. Countries, companies and individuals who fail to adopt AI will be left behind and out-competed by those who are already on board.

Humanity must bravely face the future and embrace the massive opportunities presented by Artificial Intelligence. But as we move towards that future, we cannot shy away from the issues introduced by AI. They must be tackled thoughtfully and they must be tackled head-on.

Artificial Intelligence is ultimately a technology built by humans, for humans, to make life better for us all. But whether the reality is a “for better or for worse” one, will be wholly dependent on whether we do address these fundamental and critical issues.