How AI turns citizens into auditors of power: Document forensics for everyone

Recently, a Class 12 student in India did something public institutions rarely expect of an ordinary citizen, let alone a school-going teenager. He downloaded successive versions of a public procurement tender issued by the national examination board under which he studied. Comparing the documents clause by clause, he found that the eligibility requirements had been repeatedly rewritten during the bidding process, raising questions about the integrity of the process.

His analysis reached a parliamentary committee, prompted official scrutiny, heads rolled, and demonstrated something far more significant than the procurement dispute itself. A private citizen had shown that the documents governments routinely publish can become instruments of accountability when someone has the ability to read them systematically.

The episode exposes an assumption that has quietly shaped modern democracies. Public oversight has long depended less on access to information than on the ability to process it. Procurement records, committee minutes, budget statements, disclosure filings and regulatory orders have generally been available to anyone willing to search for them. Their sheer volume, however, meant that meaningful scrutiny remained the preserve of journalists, auditors, regulators and public-interest organisations. The limiting factor was never transparency alone. It was the labour required to turn thousands of pages into evidence.

More from Opinion

Artificial intelligence changes that equation because it changes the economics of reading. Large Language Models (LLM) can compare multiple versions of lengthy documents, identify inserted or deleted clauses, trace inconsistencies across revisions and summarise patterns that previously required days or weeks of manual work. The documents themselves remain public records. What changes is the cost of extracting meaning from them. AI reduces that cost dramatically.

Governments do not always conceal information by withholding it. They often overwhelm citizens with information instead. Public procurement illustrates the problem well. A tender may be followed by corrigenda, revised eligibility conditions, technical clarifications and amended timelines. Every document is publicly available, yet understanding how the procurement evolved requires painstaking comparison across dozens of revisions. The information is open. The knowledge remains inaccessible because very few people have the time to reconstruct the sequence.

AI introduces a new form of civic capability. It allows individuals with no institutional affiliation to perform document forensics at a scale that was previously practical only for organisations. A citizen can now examine years of procurement records, compare environmental approvals across projects, identify repeated changes in regulatory language or detect inconsistencies between committee reports and final decisions. The technology does not replace legal authority or investigative judgement. It expands the range of questions that a single individual can reasonably ask.

That shift redistributes an important form of power. For much of modern administrative history, governments enjoyed a practical monopoly over their own paperwork. Citizens could request documents, but only institutions possessed the capacity to analyse them systematically. AI weakens that monopoly. The advantage increasingly shifts from those who possess the archives to those who can interpret them efficiently. Public accountability therefore becomes less dependent on organisational size and more dependent on analytical capability.

The implications extend well beyond procurement. Corporate filings, environmental clearances, municipal budgets, legislative amendments and judicial records all contain patterns that become visible only through comparison across time. Historically, these comparisons demanded specialist teams and significant financial resources. AI allows many of them to be performed by individuals using publicly available tools. The consequence is a broad expansion of who can participate in democratic oversight.

The same development also explains why governments are becoming more attentive to information governance. India's Digital Personal Data Protection Act amended Section 8(1)(j) of the Right to Information Act, changing the treatment of personal information under the disclosure framework. The government argues that the amendment strengthens privacy protections, while critics contend that it narrows access to information that previously could be disclosed in the larger public interest. The debate illustrates a broader structural tension. AI is increasing citizens' capacity to analyse public information precisely when governments are reconsidering the legal boundaries of what information should remain public.

Technology therefore cannot guarantee transparency. AI cannot analyse documents that citizens are never allowed to obtain. It cannot compensate for weak whistleblower protections, politically dependent audit institutions or disclosure systems built around scanned documents that frustrate automated analysis. These become questions of institutional design rather than technological capability.

Governments therefore face a choice. They can redesign public information systems for an era in which citizens increasingly analyse documents with AI, publishing structured and machine-readable records that make oversight more reliable. Or they can preserve bureaucratic opacity by relying on fragmented disclosures, inconsistent formats and narrower legal access. The first approach accepts that scrutiny has become widely distributed. The second attempts to preserve an administrative advantage that technology is steadily eroding.

The broader significance of AI in public life may therefore lie somewhere unexpected. Much of the public debate focuses on automation, productivity and employment. Equally important is AI's effect on democratic accountability. For the first time, ordinary citizens possess tools that substantially reduce one of the greatest practical barriers to oversight: the ability to read governments at scale.

The scarce resource in public accountability was never information itself. Governments have always generated extraordinary quantities of it. The scarce resource was the human capacity to examine those records closely enough to discover what changed, what disappeared and what deserved explanation. AI makes that capacity available to millions rather than a handful of institutions. Governance will increasingly be shaped by how states respond to citizens who can finally read the fine print.

(Sreejith Sreedharan is a technology analyst and author of 'Future of Work – AI Augmented Autonomous Decentralised'. He works on organisational AI readiness and created the AI Instinct Index®, a psychometric diagnostic designed to assess behavioural readiness for AI adoption and adaptive capacity in constraint-heavy environments. Views expressed in the above piece are personal and solely those of the author. They do not necessarily reflect Firstpost’s views.)