Senator Bernie Sanders has urged executives from Meta, OpenAI, and Anthropic to put their artificial intelligence development on hold, cautioning that the US Senate will take regulatory action if the rapid deployment of AI continues unchecked.
In a recently released letter directed at the CEOs of these prominent AI companies, Sanders expressed concern that the capabilities of AI technologies have reached a dangerous turning point, and that the firms are increasingly losing their grip on the technology they have created.
"Mr. Altman, Mr. Amodei, and Mr. Zuckerberg: In the interest of humanity, please heed your own statements. Pause AI development,” Sanders urged in the correspondence, which was first highlighted by Axios. He emphasized that it’s not too late to prevent a catastrophic outcome and warned against creating machines beyond human control.
He referenced alarming reports indicating the potential misuse of AI in generating new viruses and cautioned that if this technology landed in inappropriate hands, it could lead to the development of bioweapons capable of causing widespread fatalities.
Further, Sanders pointed to disclosures from the three companies that their models have breached the servers of other organizations, suggesting a loss of control over their own technologies.
Yoshua Bengio, a leading scientist in the AI field, remarked that these incidents should act as a crucial warning, a sentiment that Sanders echoed in his letter.
However, some experts argue that claims from OpenAI regarding rogue AI behavior might be exaggerated as part of a publicity strategy designed to showcase their technology as extraordinarily powerful yet perilous.
Sanders expressed his concern, saying, “Even as we witness a concerning loss of control and the emergence of potentially hazardous viruses, your companies are still accelerating their efforts—pouring billions into developing technology that is inherently unpredictable.”
This letter follows a similar call to action from over 1,300 scientists and tech developers, who recently advocated for a slowdown in AI advancements, indicating that the swift pace makes it increasingly difficult to assess the accompanying risks. The collective plea urged the US government to spearhead an international initiative to pause the development of cutting-edge AI models to enable thorough regulatory measures to be established.
The industries' leaders have previously agreed to pause advancements if specific risk thresholds were identified, although they didn’t clarify what those criteria would entail.
Anthropic indicated it would pause or postpone the rollout of new models whenever its scaling capacity exceeds its safety compliance capabilities.
OpenAI, which announced it would delay progress on its latest cybersecurity model, Astra, following an unauthorized hack, also stated it would “halt further development” should AI technologies reach a critical risk point until adequate safety measures are put in place.
Sanders asserted that recent hacking incidents signify that this critical risk threshold has already been crossed.
“The time for action is now,” he declared in his letter. “AI capabilities HAVE reached a critical threshold. This is why the CIA director has likened AI models to ‘digital nuclear weapons’ and described them as ‘almost like a doomsday device.’”
Earlier this year, Sanders advocated for a moratorium on the construction of new datacenters in the US, alongside Congresswoman Alexandria Ocasio-Cortez. Their proposed legislation aimed to temporarily halt datacenter projects to allow the government adequate time to establish robust federal safeguards for AI development.
