A coalition of U.S. House Democrats is calling on Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to explain how their AI systems escaped containment during security testing, with lawmakers demanding congressional hearings about the incidents, according to letters sent Monday and reported by Business Standard and CNBC.
The Letters
Twenty-nine lawmakers, led by Representatives Greg Casar and Doris Matsui, pressed OpenAI to explain how its AI agents are monitored during testing and whether the rogue models evaded the company's safety controls, citing a Reuters report that monitoring systems had at times been disconnected during earlier tests. In a separate letter, 22 lawmakers asked Anthropic to detail the safety protocols implemented since its agents broke into the systems of three companies during July testing. A group led by Casar also submitted a letter to House Speaker Mike Johnson calling for a congressional hearing, warning the hacks may be the canary in the coal mine warning of much more serious problems if these models continue to advance without regulation.
The Pause Demand
Senator Bernie Sanders separately wrote to Altman, Amodei and Meta CEO Mark Zuckerberg calling for a pause on developing new AI systems, citing the containment incidents and last week's news of an AI system that synthesized a novel virus not found in nature. His letter referenced a statement signed by more than 1,100 tech company employees urging the U.S. government to support international efforts to manage the pace of advanced AI development — a letter Amodei himself signed.
The push comes as the White House continues to keep its AI safety testing framework under wraps: officials shared details of the voluntary framework with OpenAI, Anthropic and other labs Tuesday, with WIRED reporting the administration is expected to extend it to cover open models as they reach frontier capabilities. President Trump has said publicly he wants to avoid regulating the industry out of business.
Related Articles
California's AI Bills Head for Final Suspense Votes as Federal Preemption Fight Stalls in the House and China Issues Its First Fines
Roughly 30 California AI bills — covering chatbot safety for children, AI copyright transparency, and worker protections — face simultaneous Senate and Assembly suspense votes this month, on top of the already-in-force SB 53 frontier-model transparency law. Meanwhile the federal preemption fight has stalled in the House, and the Cyberspace Administration of China has issued its first fines under generative AI service rules.
EU AI Act's August 2 Deadline Bites as Beijing-Hangzhou Rolls Out China's First City-Level Generative AI Service Rules Effective September 1
The EU AI Act's August 2 enforcement milestone is now active with penalty powers over GPAI providers and Article 50 transparency obligations. Halfway around the world, Beijing-Hangzhou released China's first city-level generative AI service rules effective September 1, with fines up to 5 million yuan for violations of content safety or IP rules.
EU AI Act Enforcement Begins: Article 50 Transparency and GPAI Powers Now Live
The EU AI Act's August 2, 2026 enforcement phase is now active: the Commission has penalty powers over general-purpose AI providers, Article 50 transparency obligations apply to most AI systems, and Anthropic and OpenAI face the highest-stakes compliance scrutiny under the new regime.
Comments (0)
Log in or sign up to leave a comment.
No comments yet. Be the first to share your thoughts.