European AI companies and policymakers pushed back against Anthropic CEO Dario Amodei’s proposal for leading developers to slow improvements in AI capabilities and jointly establish safety standards. France’s Mistral said established companies risk using safety concerns and regulation to strengthen their market position, while Proton, Black Forest Labs and Hugging Face warned that measures could restrict new entrants and open innovation. Amodei has argued that AI development is outpacing government regulation and proposed voluntary coordination among major developers, including permanent embedded evaluators from outside research organizations to assess whether companies are developing AI at a safe pace. OpenAI CEO Sam Altman has also pledged to embed outside evaluators, bringing little-known AI safety groups into greater public scrutiny. The dispute adds to the debate over whether increasingly capable AI requires slower development. Amazon supports rigorous pre-release testing, deployment thresholds and safeguards but rejects an industry-wide slowdown. Its Frontier Model Safety Framework requires models to clear specified benchmarks before release, while Amazon develops Nova models, provides third-party access through AWS Bedrock and expands AI infrastructure.