AI Regulation Should Be Rational, Not Retaliatory (Anthropic sanctions)

The Electronic Frontier Foundation filed an amicus brief arguing that the Trump administration's sanctions against Anthropic violate the First Amendment. The foundation said the measures were retaliation for the company's refusal to allow its models to be used for fully autonomous weapons or to spy on Americans. The administration designated Anthropic a supply chain risk after the company resisted those demands. The designation barred agencies and government contractors from doing business with Anthropic. A court issued a preliminary injunction blocking the sanctions from taking effect. A recent executive order imposed export controls that banned foreign nationals from using Anthropic's Mythos and Fable models. Anthropic shut the models down to comply. The administration cited concerns that the models could exploit software vulnerabilities. The foundation noted that other models with similar capabilities are subject only to a voluntary system in which companies submit them for government testing 30 days before public release. EFF said the targeted export controls on Anthropic models do not meet standards of rational, evenhanded policy. The group pointed to its past challenges to encryption export controls in the 1990s. Courts in those cases recognized software as protected speech under the First Amendment. EFF argued that similar speech concerns apply to restrictions on AI models.