WASHINGTON — Sen. Andy Kim of New Jersey is pressing Congress to establish binding rules for artificial intelligence, arguing that the companies developing the technology cannot be expected to police themselves. “The AI industry’s promise to slow itself down is not nearly enough,” Kim said Monday in a phone interview. He said lawmakers need a specific federal framework addressing the potential societal risks of increasingly powerful AI systems. Kim, a Democrat who serves on the Senate Commerce Committee, said outside evaluators reviewing AI laboratories could be part of an oversight system but would not be sufficient on their own. Anthropic recently said it would allow independent evaluators access to its labs. “We really need a very specific regulatory framework when it comes to AI,” Kim said. “Yes, embedding evaluators is certainly a part of that, but that is not going to be enough to address my concerns about where this technology is headed.” He added that Congress must define the standards evaluators would use and the responsibilities companies would have to meet. Relying on what he called the “voluntary goodwill” of AI companies, he said, would leave the public dependent on corporate promises while the technology continues to advance. Kim’s remarks followed warnings from AI researchers and industry leaders about the consequences of unchecked development. Three major AI executives — Dario Amodei of Anthropic, Sam Altman of OpenAI and Elon Musk of xAI — said over the weekend that mechanisms are needed to slow the pace of progress and reduce the risk of catastrophic outcomes. The New Jersey senator also referred to Anthropic’s powerful “Mythos” model and said society needs to be “Mythos-proof.” He said he has been working with multiple parts of government on that issue but did not provide details about a specific proposal. The debate is unfolding as lawmakers face pressure over both AI safety and the rapid construction of data centers needed to power the industry. A bipartisan Senate effort to advance AI safety legislation remains stalled, with disagreements over provisions that could make AI laboratories legally accountable. Kim said he has had productive bipartisan discussions and hopes the Senate Commerce Committee will hold a markup in the coming weeks. President Donald Trump has argued that additional safeguards could undermine the United States’ position in its technological competition with China. In a Monday post on Truth Social, Trump said AI needs the control of a “STRONG AND SMART” president rather than new guardrails. Kim rejected that framing, saying regulation and technological competitiveness are not opposing goals. “Those are not mutually exclusive,” he said. “We can find bipartisan support to show that there is a real and better way that we can do this.” Congressional leaders have not committed to moving quickly. House Speaker Mike Johnson said a White House meeting with executives from major AI companies could take place later this week or early next week, potentially with Trump attending. Johnson has emphasized corporate self-regulation while acknowledging that Congress may have a role. The House is also preparing to consider legislation requiring technology companies to absorb more of the rising energy costs associated with data centers. The issue has gained urgency as communities and voters weigh the effects of new facilities on electricity demand, infrastructure and utility bills. With lawmakers facing a narrow schedule before the midterm election recess, the push for federal oversight remains politically unsettled. Kim’s position reflects a growing argument among lawmakers that public protections should be set through enforceable rules rather than left to companies whose profits depend on expanding AI development.