AI 'kill switch' bill unveiled as OpenAI hack raises alarms

July 24, 2026 - 00:54

A bipartisan group of House lawmakers is set to introduce a bill on Thursday that would give the federal government the power to shut down or restrict artificial intelligence systems deemed too dangerous. The legislation, which has been dubbed the "kill switch" bill, arrives just days after a security breach at OpenAI raised fresh concerns about the safety of advanced AI models.

The proposed law would authorize a new federal oversight body to monitor AI development and issue emergency orders to halt operations of models that pose an imminent threat to national security, public safety, or critical infrastructure. Under the bill, companies developing powerful AI systems would be required to report safety incidents and submit to regular audits. Failure to comply could result in fines or forced shutdowns.

The timing is notable. Reports emerged this week that hackers accessed internal communications at OpenAI, the company behind ChatGPT. While the breach did not compromise customer data or the core AI model itself, it exposed vulnerabilities in the company's security protocols. Lawmakers cited the incident as a clear example of why regulatory guardrails are necessary.

Supporters of the bill argue that the current voluntary safety commitments from tech firms are not enough. Critics, however, warn that giving the government a kill switch could stifle innovation and hand too much control to bureaucrats who may not fully understand the technology. The bill faces an uncertain path in a divided Congress, but its introduction signals that Washington is taking the risks of unchecked AI seriously.