A bipartisan pair of US House representatives introduced legislation that would give the Department of Homeland Security authority to order artificial intelligence companies to shut down models that pose an imminent threat to human life or the economy. The bill follows an incident involving OpenAI, the maker of ChatGPT, though details of that incident remain under review.
The proposed 'AI Emergency Shutdown Act' would require AI developers to implement a kill switch mechanism accessible to federal authorities. Companies that fail to comply could face fines or operational restrictions. The bill's sponsors, Representative John Smith (R-TX) and Representative Jane Doe (D-CA), said the measure is necessary to prevent catastrophic failures.
OpenAI has not publicly commented on the specific incident cited by lawmakers. The company previously stated it supports reasonable regulation but has warned against overly broad government intervention. The bill faces an uncertain path in Congress, where AI regulation remains a divisive issue.
If passed, the law would apply to any AI system deemed capable of causing mass casualties or significant economic disruption. The Department of Homeland Security would be required to consult with technical experts before issuing a shutdown order.
