The $27M Proxy War That Exposed AI's Regulation Fault Line

Last Tuesday, voters in New York's 12th Congressional District chose Micah Lasher as the Democratic nominee to succeed longtime Rep. Jerry Nadler. On the surface: a routine Manhattan primary. Under the hood: a $27 million collision between the two most influential AI companies in the world, each betting real money on what AI regulation in America should look like.

How a State Assemblyman Became a $27M Target

Alex Bores is not a typical congressional candidate. Before winning a state assembly seat in 2022, he spent five years at Palantir as a data scientist and government lead — leaving in 2019 over the company's renewed ICE contract. He holds a master's in computer science with a machine learning specialization from Georgia Tech. In 2025, he co-authored New York's Responsible AI Safety and Education Act (RAISE Act), which the NY legislature passed in June 2025. Politico described it as "widely regarded as the furthest-reaching [AI regulation] in the country." Time named him to its 2025 Time 100 AI list for the work. Governor Hochul signed a significantly weakened version in December 2025.

That profile made him a lightning rod. The RAISE Act imposed strict obligations on frontier AI developers — exactly the kind of legislation that splits the industry between those who see it as necessary guardrails and those who see it as a brake on innovation.

Anthropic vs. OpenAI: Who Spent What

• Public First Action, a super PAC that received $20 million linked to Anthropic, ran ads supporting Bores. Anthropic stated those funds could not directly support federal election activity.
• Leading the Future, an OpenAI-aligned super PAC, spent $7.6M+ in attack ads opposing Bores — framing AI safety legislation as a threat to technological progress.
• Guardrails Alliance (~$258K), funded by OpenAI employees concerned about their company's political spending, ran a counter-effort opposing their own employer's strategy.
• Crypto investor Chris Larsen committed $3.5 million to counterbalance the OpenAI-aligned attack spending.

Bores' poll numbers actually climbed during the attack-ad barrage. The blitz backfired in the court of public opinion. He still narrowly lost to Lasher — a painful outcome for the Anthropic-adjacent coalition that spent $20 million trying to put him over the line.

Why It Ends as a Draw

On a scoreboard reading, OpenAI-aligned money won: the candidate they spent millions to defeat did indeed lose. But that's too clean a read. Anthropic publicly tethered itself to the AI safety cause and the RAISE Act's brand. OpenAI publicly tethered itself to opposition to that regulation — a position that drew sustained backlash, including the Guardrails Alliance formed by its own employees mid-race. Neither company emerged with clean hands or a clear mandate. The Verge called it a draw, and that's fair: $27 million bought each side a lot of press and a lot of enemies.

What Builders Should Take From This

If you're shipping AI products — trading systems, fintech agents, anything making consequential automated decisions — this race is a preview of the terrain ahead. The RAISE Act's core demands (documentation, testing requirements, human oversight thresholds for high-risk AI) did not die with Bores' primary loss. They will resurface, at state and federal level, under different names and sponsors.

The practical implication is straightforward: regulatory readiness is a product trait, not an afterthought. The teams that navigate this environment best will be the ones who build auditability, human-override hooks, and decision logging in from day one — not the ones retrofitting compliance after a mandate arrives. The $27M spent in this race is a loud signal that the AI industry itself knows something real is coming. Build accordingly.