Congress Scrambles on AI Preemption; Anthropic Fights to Lift Foreign Model Ban

Congress scrambles to pass AI preemption before midterms as Anthropic opens talks to lift its government-imposed foreign model ban.

🏛️ Big Tech Makes Its Final Push for Federal AI Preemption — Bundled With Child Safety

Decoded: With midterm elections looming, Big Tech lobbyists are making a final push for federal AI preemption — a single national law that would override the patchwork of state-level AI regulations adopted across the U.S. The White House this week proposed bundling the preemption measure with the Kids Online Safety Act (KOSA), backed by Sen. Marsha Blackburn (R-TN), creating a package that combined two of the most contested tech policy fights in Congress. The move created immediate friction: House Republicans said they had not been consulted, Democratic KOSA co-sponsors said they were left out of planning, and a separate bipartisan AI preemption bill was already circulating in the House. "No one knows really who's driving this thing," a Republican lobbyist for a midsize tech company told The Verge. President Trump has publicly called for AI preemption to pass, putting Republican leadership in a difficult position as the party's populist bloc remains skeptical of legislation seen as protecting Big Tech incumbents. Passage before a potential House flip to Democrats after the midterms is the industry's stated deadline. (The Verge, June 15, 2026)

Why it matters: Federal AI preemption is the single most consequential legislative outcome for U.S. AI companies. Without a national law, AI developers face the legal and compliance cost of navigating diverging regulations in more than a dozen states — California, Texas, Colorado, and New York have each advanced their own frameworks, with conflicting definitions of high-risk AI, liability exposure, and disclosure requirements. For Microsoft (MSFT), Google (GOOGL), and Meta (META), preemption would replace that regulatory maze with one federal standard, reducing compliance drag and eliminating the risk that any state can impose more restrictive rules on nationally deployed products. The KOSA bundling creates a political wildcard: child safety legislation has broad public support but deep intraparty complexity over which version advances, and failure to resolve internal Republican disputes before the midterm calendar closes could kill both bills simultaneously.

🤖 Anthropic Opens Negotiations to Restore Foreign Access to Fable 5 and Mythos 5

Decoded: Anthropic has entered formal talks with U.S. government officials to lift the export control directive issued June 13 that required the company to disable its Fable 5 and Mythos 5 AI models for all foreign nationals. The company is contesting both the evidentiary basis and proportionality of the ban, which cited a claimed jailbreak vulnerability allowing the model to identify software weaknesses in critical infrastructure. Anthropic said publicly that the government provided only "verbal evidence of a potential narrow, non-universal jailbreak" and that the order did not follow principles of "fair and fact-based regulation." The company's pre-IPO timeline adds pressure: the longer the restriction persists, the more it constitutes a material disclosed risk in the S-1, capping Fable 5 international revenue at zero during the roadshow period. (Reuters, June 16, 2026)

Why it matters: The outcome of Anthropic's negotiations will establish how export control directives targeting deployed AI software can be challenged or modified — entirely new legal terrain since the June 13 directive was the first time the U.S. government ordered a commercial AI lab to restrict access to a live model on national security grounds. If Anthropic succeeds in narrowing or lifting the ban, it creates a contestation precedent that reduces tail risk of unilateral government restriction across the sector. If the ban holds through the IPO roadshow, it becomes a disclosed material risk that institutional investors will price into the Anthropic valuation. Google (GOOGL), which holds a major Anthropic equity stake, carries secondary exposure to the IPO valuation outcome in either scenario.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team