Council says AI behind increase in FOI requests

Lincolnshire County Council said AI-generated Freedom of Information requests were partly behind an 18% increase, taking total FOI requests to 1,992 in the 2025/26 financial year. The council also missed its 85% response-time target, with 83% of replies issued within 20 working days. Management is now assessing whether AI can also help speed up responses, although staff will still review outputs for accuracy.

Analysis

This is less a story about public-sector paperwork and more an early signal that low-friction, high-volume natural-language generation is becoming a workload amplifier for compliance-heavy organizations. If AI is now shaping both the demand side and the supply side of administrative processes, the first-order effect is not just more requests — it is a step-change in labor intensity, QA burden, and response-liability risk across local government, universities, utilities, and other FOI-exposed entities. The beneficiary set is likely to be vendors selling workflow orchestration, records management, e-discovery, and human-in-the-loop review rather than generic LLM providers.

The second-order pressure point is governance: once councils start using AI to draft responses, the risk profile shifts from backlog to mis-disclosure, exemption errors, and appeal amplification. That creates a paradoxical demand loop: faster drafting can increase throughput, but even a small rise in defective responses can trigger more complaints, ombudsman scrutiny, and legal review costs over the next 6-18 months. In other words, the near-term productivity gain may be offset by a longer-duration control stack build-out.

The market is probably underestimating how quickly procurement budgets can reallocate toward vertical AI tools with audit trails, citation checking, and policy enforcement. The winning model is not “chatbot for citizens,” but internal copilots that sit on top of case management systems and preserve evidentiary logs. This favors software platforms that can prove defensibility, not just speed; it is a classic enterprise AI adoption pattern where trust becomes the bottleneck after the pilot phase.

Contrarian view: the headline may overstate the novelty of AI and understate the broader procedural inflation in information rights requests. If request volume is structurally driven by transparency activism and templated campaigns, the right response is process redesign and filtering, not more headcount or a pure LLM layer. That means the strongest trade is against vendors exposed to simplistic automation claims and in favor of companies that monetize governance, retrieval, and compliance validation.

Key Decisions for Investors

• Initiate a 3-6 month long position in MSFT versus a basket of generic AI names (or QQQ) if you want exposure to workflow AI monetization with lower execution risk; Microsoft benefits more from embedded copilots, auditability, and enterprise distribution than standalone model hype.
• Add to positions in GOOGL and/or AMZN on pullbacks, with a 6-12 month horizon, as governments and regulated enterprises increasingly favor cloud-native retrieval and governance tooling; use this as a selective long, not a broad AI-beta trade.
• Short small-cap "AI automation" software names with weak governance or vertical specificity on any post-headline rally; the risk/reward is attractive over 1-3 months because procurement buyers tend to reject black-box systems after the first compliance incident.
• Pair trade: long enterprise workflow/compliance software (e.g., DOCU, PATH, or similar governance-adjacent software where applicable) versus short pure-play LLM monetization proxies; thesis is that audit trails and human review will be the durable budget line item over the next 12 months.
• Watch for a 1-2 quarter lag before local-government AI procurement shows up in vendor commentary; use that as a catalyst window to add to names with records-management and case-workflow exposure, not consumer-facing AI.