House AI ‘kill switch’ bill unveiled as OpenAI hack raises alarms - Politico

What the report says

Politico reported that a House proposal on artificial intelligence would give the federal government authority to intervene against AI systems deemed dangerous, including by ordering a model taken offline or limiting how it operates. According to the outlet’s headline and available snippet, the bill was introduced by the co-chair of a major Democratic House AI panel together with a Republican lawmaker, signaling at least some cross-party interest in stronger safeguards for advanced AI.

The report, dated July 23, 2026 in Politico’s URL, frames the legislation against concerns raised by an OpenAI hack. The accessible article text does not identify the lawmakers, describe the alleged hack in detail, or specify which agency would receive the proposed authority, so those points remain unclear from the available source material.

If advanced by Congress, the measure could become part of a broader debate over how much emergency power the government should have over privately developed AI systems. Supporters of such powers generally argue that fast-moving AI risks may require rapid intervention, while critics often warn that broad shutdown authority could affect innovation, competition, civil liberties or the reliability of tools used by businesses and the public.

The bill matters because it points to a more forceful regulatory approach than voluntary safety commitments or post-release oversight. Even without full legislative details, Politico’s report indicates lawmakers are considering whether high-risk AI models should be subject to direct federal controls when security or public-safety concerns arise.