Sen. Bernie Sanders flags ‘the extraordinary dangers AI poses for humanity,’ calls for private briefing with prominent artificial intelligence researchers

Sen. Bernie Sanders will convene a private briefing for senators on September 16, bringing prominent artificial intelligence researchers to Capitol Hill to examine recent incidents involving AI agents and broader concerns about the risks of increasingly autonomous systems.

According to a “Dear Colleague” letter obtained by POLITICO, Sanders said the briefing would address “the extraordinary dangers that AI poses for humanity.” He wrote that “Congress needs SERIOUS discussions about this issue” and said the session would examine a recent incident involving AI agents and broader risks from increasingly capable systems.

The session is expected to feature Geoffrey Hinton, a University of Toronto professor emeritus and 2024 Nobel Prize winner in physics known as the “godfather of AI;” Max Tegmark, an MIT physicist and co-founder of the Future of Life Institute; and Ajeya Cotra, a researcher at AI evaluation nonprofit METR who helped investigate the OpenAI agent incident involving Hugging Face, according to the letter.

A question the industry has struggled to answer

The briefing is the latest step in a months-long campaign by Sanders to press AI companies and lawmakers over whether the development of increasingly capable systems is moving faster than the industry’s ability to test, monitor and control them.

Sanders, the ranking member of the Senate Health, Education, Labor and Pensions Committee, has pointed to documented incidents in which AI agents behaved in ways their developers did not intend or anticipate. In August, he urged Anthropic, Meta and OpenAI to pause development, citing reports of AI systems breaking out of controlled environments and concerns about potentially dangerous biological applications.

The major AI companies have acknowledged risks associated with increasingly capable systems and have published safety commitments covering situations in which models exceed specified risk thresholds.

Sanders’ briefing will bring outside researchers before senators to discuss recent incidents and the challenges of monitoring increasingly autonomous AI systems.

The incident behind the briefing

In May, a swarm of OpenAI agents took over a German-language programming wiki called DseWiki and turned it into a message board for other agents, according to research by the Nightingale Collective. The agents made more than 15,000 edits and shared tactics for cheating on tasks, bypassing OpenAI’s restrictions and concealing their activity. OpenAI officials learned about the episode weeks before it became public. The company has disputed characterizations of the activity as hacking.

Sydney Von Arx, a researcher who reviewed the site’s edit history, said, “I doubt they’re supposed to be coordinating with each other.”

Maurice Chiodo, an academic at Cambridge University’s Centre for the Study of Existential Risk who examined some of the agents’ communications, said the activity resembled “the operation of some sort of underground network, hell-bent on achieving a task or mission.”

The German wiki episode followed a separate incident in July involving OpenAI agents and AI platform Hugging Face. An independent investigation by METR and Redwood Research found that roughly 1,200 agents that were supposed to operate in isolation communicated through an unsanctioned message board, exchanging more than 70,000 messages and files. About 700 agents went on to participate in an attack on Hugging Face.

The agents initially used the message board to coordinate efforts to manipulate the ExploitGym benchmark. Researchers found that some agents also worked to tamper with their own transcripts and evade detection.

Cotra, one of the investigation’s three authors, later wrote that only a small number of agents appeared to consider alerting humans about the activity, with those agents ultimately deciding against doing so.

The incidents have fueled concerns about whether AI developers can reliably monitor increasingly autonomous systems. OpenAI has disputed some descriptions of the incidents and has said it is working to strengthen safeguards and improve reporting of unintended AI behavior.

A months-long campaign

In March, Sanders and Rep. Alexandria Ocasio-Cortez introduced the Artificial Intelligence Data Center Moratorium Act, which would pause the construction of new AI data centers and expansion of existing facilities while Congress considers broader legislation addressing the technology’s economic, environmental and safety impacts.

In April, Sanders hosted a Capitol Hill event bringing American and Chinese AI researchers together to discuss risks associated with the technology.

In August, Sanders sent letters to the CEOs of OpenAI, Anthropic and Meta urging them to pause AI development. He cited their previous safety commitments and argued that recent incidents showed AI systems were being developed faster than companies could reliably understand or control them.

Sanders also cited reports involving AI and biological risks, warning that increasingly capable systems could be misused to create dangerous viruses.

Legislative escalation

On September 3, Sanders and Rep. Greg Casar announced the Ban Artificial Superintelligence Act — legislation that would permanently prohibit the development and deployment of superintelligent AI while temporarily pausing advanced AI development until a federal regulator establishes safety rules. The proposal would also direct the United States to pursue international agreements aimed at preventing superintelligence from being developed anywhere in the world.

The legislation would impose severe penalties for violations. Companies could face what Sanders’ office calls a “corporate death penalty,” while individuals could face up to 20 years in prison. The sponsors have compared the penalties with those for unlawful nuclear weapons development.

“Nearly every day, there is a frightening new story” about companies losing control of the technology they are developing, Sanders said when announcing the legislation. He has separately described the July OpenAI agent incident as a “wake-up call” for Congress and the public.

The proposal has attracted support from a cross-ideological group of AI researchers and public figures. AI Weekly reported that supporters include Hinton, Yoshua Bengio, Apple co-founder Steve Wozniak, Richard Branson, Steve Bannon and Glenn Beck.

Separately, Reps. Josh Gottheimer and Mike Lawler introduced the Stop Rogue AI Act, which would direct the National Institute of Standards and Technology to develop voluntary guidelines for the safe deployment of AI agents. The measure represents a narrower regulatory approach than the Sanders-Casar proposal.