Anthropic’s Mythos model is scaring House Homeland Security Committee Chair Andrew Garbarino (R-N.Y.).
Garbarino told Punchbowl News on “Fly Out Day” that during a closed-door demonstration, Anthropic showed members that the new AI model could wipe out private bank accounts.
Anthropic “told the model to find a vulnerability in a bank and empty accounts, and then it went and did it,” Garbarino said. Mythos “then could find this vulnerability and fix it.”
The demonstrations from Anthropic and other artificial intelligence companies have highlighted the need for the government to “act soon” to get a grip on the threats posed by AI, Garbarino said.
Yet Garbarino admitted that acting to address this issue is easier said than done for Congress. “I say act, but I don’t know what the answer is,” Garbarino said. “The real problem here is we don’t know what the answer is, we don’t know what we have to do.”
One thing Garbarino said he’s certain of is that federal agencies and Congress should have their hands on these platforms before they’re released.
Garbarino also said he “appreciates” Reps. Jay Obernolte (R-Calif.) and Lori Trahan’s (D-Mass.) recent discussion draft to set up a federal framework for AI.
Bad actors could hack the models and use them to attack critical infrastructure, he said, noting that AI’s agentic capabilities would allow models to take action without human prompting.
“You have the ability to not just learn how to do it, but to tell it to do it for you,” Garbarino warned.