The New Frontier of Accountability? Monitoring AI Usage in Government

By Gabriel Šípoš, Senior Legal Manager, Governance and Accountability

AI-generated image created by CoPilot

As governments around the world pour tens of billions of pounds into AI-based systems every year, civil society needs to prepare for a new challenge: how to monitor the procurement and deployment of AI tools at a time of diminishing civic space and shrinking donor funding.

AI‑based technologies have undoubtedly opened new, more efficient ways to tackle corruption, an area that represents a significant part of our work at International Lawyers Project (ILP). In two major anti-corruption conferences which ILP attended in recent months - the OECD Anti-Corruption and Integrity Forum in Paris and the Conference of State Parties to the UN Convention against Corruption in Doha - not a day passed without governments showcasing their flashy new AI products detecting corruption red flags in procurement, social services provision or internal government processes. “We will seize the opportunities presented by artificial intelligence,” states the UK government in its Anti-Corruption strategy revealed in December 2025, promising to roll out a prototype of an AI-based corruption investigation assistant next year.

But at the same time, many concerns emerge that these tools may not work as intended or may even be intentionally misused. Various benefit fraud scanning and facial recognition algorithms have failed spectacularly in various countries, the UK included. A recent report by the Institute of Development Studies pointed out that AI-based systems, meant to fight crime, are also being used to crack down on legitimate opposition activities.

Adjusting the Anti-Corruption Toolkit

Many traditional approaches used by good governance watchdogs to promote accountability, such as encouraging public participation or demanding transparency of government data and decisions, are less effective when applied to AI. Proprietary, patent-heavy technologies allow for far less contract transparency than when looking into utilities or housing procurement, for instance. The AI technologies themselves are more difficult to assess: unlike, say, more traditional forms of government spending, such as hospital or road construction, AI tools are intangible and complex, making community monitoring nearly impossible. You can’t easily see if AI was implemented, is working as planned, and is good value. In contrast, hospital clients or motorway drivers can detect faulty products almost immediately.

AI tools rely on large volumes of private data, so independent monitoring of their impact is also hard to conduct. Yet they are multi-million projects that will affect a large share of the population - precisely the kind of public spending that requires more scrutiny, not less. Moreover, the balance of power is skewed: supplier competition is limited, with a handful of powerful AI companies holding far more influence than many governments.

These dynamics are often observed in the defence and IT sectors, two areas historically associated with high corruption risks, where civil society has made only limited progress in increasing oversight.

Building Capacity for the AI Age

Despite the challenges, there are still approaches that civil society can use, but they require strengthening and adaptation. Support for whistleblowers remains essential. Their work is especially valuable where secrecy, privacy concerns and limited market competition make external scrutiny of government and their suppliers nearly impossible. A number of recent revelations about AI misuse came from insiders. More investment is needed to protect and, where necessary, legally defend whistleblowers so they feel secure when exposing abuses of power.

Other traditional tools will need to be recalibrated. Freedom of information requests can help establish whether AI tools are even being used. Last year, the UK courts confirmed that citizens have the right to know whether a public agency uses AI in its decision-making in many circumstances. But the claimant had to go to court to secure this outcome. AI‑related information requests are likely to require more precise reasoning to overcome arguments related to commercial confidentiality, privacy or even national security.

Similarly, monitoring AI procurement will require new skills, including understanding best practice terms of reference, identifying risky contract loopholes, analysing pricing benchmarks and evaluating data‑use clauses. This will require specialist expertise, from lawyers to AI product analysts. Ideally, civil society will also need training in conducting or reviewing algorithmic impact assessments, which evaluate the risks of AI systems before deployment.

 AI-generated image created by CoPilot

As many countries begin drafting their first AI regulations, civil society must also be able to scrutinise these draft laws, supported by best‑practice analyses. Striking the right balance between using AI to improve efficiency and preventing its harms and abuses will be impossible without adequate legal capacity.

Needless to say, civil society organisations and journalists will also continue to need protection from legal harassment, including SLAPPs. They will require training and sustained funding to monitor AI procurement and deployment effectively. And clearly, none of these recalibrated tools can be deployed without sufficient capacity. This is no small ask at a time when many media outlets and NGOs are struggling to survive due to shrinking civic space, limited resources, or both. Pro bono legal support to draft those information requests or appeals, provide advice on legal routes to scrutinise controversial AI projects or provide legal protection to whistleblowers, can serve as a vital lifeline: and ILP can facilitate much of that assistance. Building cross-sectoral and international collaborations to share expertise and best practices is another important step.

Ultimately, it is becoming increasingly clear that AI technologies require robust, properly resourced oversight - and urgently so.

International Lawyers Project already provides pro bono legal support to more than 100 civil society organisations, journalists and whistleblowers each year.

If you are working on AI procurement or corruption‑related AI issues, or plan to do so, please contact us at contact@internationallawyersproject.org and we can help connect you with the legal assistance you may need.