House Democrats spent Wednesday drawing two lines around AI: one around its money, one around its safety record. With the midterms days away and the AI buildout financed by a growing stack of debt and guarantees, Democrats are staking out the position that the public will not be the backstop — and that the labs will not grade their own homework.

The first line is a resolution introduced by Reps. Greg Casar of Texas and Maxine Waters of California, according to the Daily Caller. It calls on Congress to oppose any rescue of insolvent AI-related firms or any shield for executives over "reckless conduct that contributed to the failures of such firms." Failing companies should wind down through ordinary bankruptcy, it says, while executives who took excessive risks or broke the law should be "held fully accountable under civil and criminal law."

Be clear about what this is: a simple House resolution expresses sentiment, not law, and needs no Senate or presidential approval. Its power is political, a marker laid down before a crisis. S&P estimates put collective AI capital spending by Amazon, Microsoft, Alphabet, Oracle, Meta, and SpaceX above $7 trillion from 2025 to 2030, with tech borrowing projected to hit a record $420 billion in 2027, up 60% from 2026.

The second line came the same day from the ranking Democrats on three committees with AI jurisdiction — Frank Pallone Jr., Zoe Lofgren, and Jamie Raskin. In letters to Anthropic, Google, Meta, OpenAI, and SpaceXSI, they demanded a full accounting of identified misalignment and safety incidents by November 11, plus notice within 48 hours of any new incident — 24 hours if critical infrastructure is involved. Voluntary disclosure, they argue, "relies too heavily on the consistent goodwill of all AI companies to voluntarily disclose incidents that might expose them to liability and reputational damage."

Together, the two moves are one argument: no to the upside of moving fast plus the downside protection of being too important to fail. The politics favor them — an October 21 ABC News/Washington Post/Ipsos poll found 57% of Americans believe AI is bad for society, and an AP-NORC survey found 64% say it is developing too quickly.

What to watch: November 11, when the five labs' accountings are due, and whether the no-bailout language gains Republican signatures — a pledge only bites if it survives the moment a company actually fails.

— via Daily Caller