AI-powered nimbyism could grind UK planning system to a halt, experts warn (www.theguardian.com)

UK startups and community groups are using large language models to automate objections to planning applications, creating a potential counterforce to government plans that use AI to speed housing approvals. Services such as Objector (£45 per objection) and Planningobjection.com (£99) scan application documents, rank legal or policy issues as “high/medium/low”, and generate objection letters, committee speeches and even videos. Founders say the tools democratise access to planning expertise; critics — including planning lawyers — report AI outputs that cite fabricated case law and warn a deluge of polished, machine‑written submissions could overwhelm planning officials and delay or distort decision-making. For the AI/ML community this is a clear example of dual‑use LLM deployment with systemic consequences. Key technical details: the tools rely on generative models prone to “hallucination,” and Objector attempts mitigation by cross-checking outputs from two models. The story highlights risks around provenance, hallucination detection, and scale (automated mass submissions), and suggests an emerging “AI arms race” between pro‑development and anti‑development actors. Governments are already preparing — deploying their own tools (Extract, Consult) to process responses — but the situation underlines the need for robust verification, model auditing, and provenance tooling if LLMs are to be safely integrated into public‑sector workflows.