Anthropic's Leaked AI Bill Could Let Courts Block Dangerous AI

A draft AI bill, reportedly being pushed by Anthropic, has leaked. It proposes a new way for the federal government to block AI systems that could cause catastrophic harm. Right now, courts can't easily step in to stop an AI lab from releasing a dangerous model. This bill would give the Commerce Secretary the power to seek a court order to halt such actions. The idea is to prevent disasters before they happen.

The bill also deals with state laws. It would override some state AI regulations, like those in New York and California, but only if the federal government has its own blocking mechanism in place. This is different from earlier attempts to simply freeze state AI laws. The drafters seem to want a federal system that can act, not just deregulate.

For everyday people, this could mean more oversight of powerful AI. If an AI system threatens safety, the government could step in faster. But it also means states might lose some ability to regulate AI on their own. The bill includes safeguards: blocking orders expire after 45 days, and courts must consider less restrictive options first. Still, it's a significant shift in how AI is governed.

Some experts worry the bill might not go far enough or could be misused. Others think it's a necessary step. The leak has sparked debate about who should control AI risks—courts, Congress, or states. As AI becomes more integrated into daily life, these decisions will affect everything from job security to privacy. The bill's fate will shape how safe AI is for all of us.