AI Governance Becomes Political
- Recent videos argue AI regulation is moving from abstract safety debates to political fights over market access and procurement. - One piece framed the rising 'insider threat' from internal model misuse, while another examined industry reactions to vocal critics like Alex Bores. - The tone shift pushes AI governance toward operational responsibilities across security, procurement and legal teams rather than pure policy talks ((youtube.com)).
AI governance is moving out of safety seminars and into elections, procurement fights and internal security rules. (politico.com) The clearest political example is in New York, where Assemblymember Alex Bores helped sponsor the RAISE Act and then became a target of industry-backed campaign spending during his 2026 congressional run. Governor Kathy Hochul signed the law on December 19, 2025, after the Assembly passed it on June 13 and the Senate passed it on June 12. (governor.ny.gov) The RAISE Act requires large frontier-model developers to publish safety protocols, use third-party review and report qualifying incidents to New York within 72 hours after determining one occurred. The law also creates an oversight office in the Department of Financial Services. (nysenate.gov) That turns regulation into a market-access question. If a company wants to sell powerful AI systems in New York, it now has to meet concrete reporting and governance rules, not just make voluntary promises. (governor.ny.gov) The campaign money around Bores shows how fast that shift has become electoral. TechCrunch reported on March 3 that the super PAC Leading the Future had raised $125 million to oppose candidates pushing tougher AI rules and support lighter-touch regulation. (techcrunch.com) CNBC reported on January 30 that Leading the Future’s contributors included Greg Brockman, Joe Lonsdale, Ron Conway, Andreessen Horowitz and Perplexity. Politico later reported that connected AI-linked PACs were already spending in primaries in Texas, Illinois and New York. (cnbc.com) (politico.com) At the same time, the governance debate has moved inside companies. METR wrote in December 2025 that frontier AI safety policies now center on evaluations, deployment safeguards, information security controls and named accountability practices. (metr.org) That is where the “insider threat” framing comes in. RAND’s 2024 report on frontier-model security focused on protecting model weights — the files that encode a model’s capabilities — from theft and misuse, while newer security guidance treats employee use of external AI tools as a route for leaking code, customer data and strategic plans. (rand.org) (cyberhaven.com) Legal and compliance teams are being pulled in alongside security teams. EY said in September 2025 that AI governance was becoming a practical compliance function, and Bloomberg Law said last week that companies now need documented policies, device controls and audit-ready records. (ey.com) (bloomberglaw.com) The result is a different kind of AI politics. The fights are no longer only about distant existential risk; they are about who can buy, sell, deploy, insure and campaign around AI systems under rules that now carry deadlines, disclosures and enforcement. (nysenate.gov) (politico.com)
