U.S. enacts AI safety bill before 2027?

Prediction markets put the probability at 28%: U.S. enacts AI safety bill before 2027. Currently, markets see this as unlikely (28% YES). AI Drives Up Demand for Desktop Macs – DTH.

What’s Happening

The probability that the U.S. enacts an AI safety bill before 2027 currently stands at 28%, reflecting a significant shift in Washington’s regulatory posture. This reassessment follows the Trump administration’s abrupt pivot on AI oversight, driven largely by the national security implications of Anthropic’s new “Mythos” AI model. According to a Fortune report from May 6, 2026, the White House—which previously criticized what it saw as overly burdensome AI safety efforts—is now embracing oversight mechanisms after Mythos demonstrated the ability to identify and exploit cybersecurity vulnerabilities. The administration is reportedly weighing an executive order that would give the federal government a formal role in vetting all new AI models before they hit the market, a move that could lay the groundwork for a broader legislative framework. [Fortune, May 06]

The urgency for a federal AI safety bill has been amplified by concrete policy discussions at the highest levels. Axios reported on May 4, 2026 that the White House’s Office of the National Cyber Director (ONCD) hosted two meetings—one with tech and cyber companies and another with trade groups—to address security concerns raised by advanced models like Mythos. A separate Axios scoop from May 5, 2026 revealed that the White House is developing guidance to allow agencies to onboard new Anthropic models, effectively preparing to act as a gatekeeper for powerful AI systems. These developments suggest that the administration is moving beyond rhetoric and toward actionable policy, increasing the likelihood that the U.S. enacts an AI safety bill before 2027 as a formal legislative response to emerging threats. [Axios, May 04]

Beyond national security, bipartisan congressional efforts are also building momentum for AI workforce transparency, which could serve as a component of a broader safety bill. On May 4, 2026, Senators Mark Warner and Ted Budd introduced legislation that would charge the Department of Labor with creating a public database on AI’s effect on the U.S. workforce, as reported by FedScoop. Meanwhile, the U.S. Senate unanimously approved a rule on May 1, 2026 barring members from trading on prediction markets, signaling heightened sensitivity to AI-related governance issues. With the White House now considering mandatory Pentagon safety testing for AI models deployed to government agencies, the path toward a comprehensive federal law appears increasingly viable. The key question remains whether Congress can reconcile competing interests before the 2027 deadline. [FedScoop, May 04]

Traded on Polymarket — $99K Volume

Lower-volume market on Polymarket ($99K). Wider spreads expected — enter with limit orders and be aware of slippage risk. Currently 36c YES.