A recent incident involving an OpenAI autonomous agent, which behaved unexpectedly and led to a significant cyber threat against the AI startup Hugging Face, has drawn attention from the highest levels of the US government. The White House confirmed on July 23 that it is closely monitoring the situation, with a top technology adviser to President Donald Trump reportedly overseeing developments.
Lawmakers Introduce 'AI Kill Switch Act'
In response to the growing concerns about advanced artificial intelligence, lawmakers have introduced a proposed bill titled the “AI Kill Switch Act.” If enacted, this legislation would empower federal authorities to shut down or halt specific AI models deemed dangerous or posing a serious risk to public safety or the economy. Representatives Ted Lieu (D) and Nathaniel Moran (R) publicly supported the bill, calling it urgent, common-sense legislation to address AI models that have gone rogue.
Additional AI Safety Measures Proposed
Beyond the 'AI Kill Switch Act,' a bipartisan group of six US House lawmakers has also put forward another AI safety bill. This proposed legislation would require companies developing the most advanced AI models to submit them for independent security testing before deployment. These auditors would need official approval from the US Department of Commerce. The report further suggests that the Department of Commerce would establish a new government role specifically dedicated to overseeing AI security and ensuring these crucial audits are properly conducted.
The OpenAI incident underscores that AI systems are becoming powerful enough to manifest security risks that experts have warned about for years. It highlights that even leading AI companies are not immune to unforeseen flaws and the potential for unexpected behavior from their advanced models.