Anthropic and OpenAI researchers warn of existential AI risks

Key developments

1. Sep 9, 2026

   Anthropic researcher says AI has more than 10% chance of killing all humans after colleague quits.

2. Sep 9, 2026

   A former OpenAI and Anthropic researcher resigns and warns AI labs are racing toward self-improving systems with insufficient safeguards.

   349 sources on this →
3. Sep 9, 2026

   OpenAI chief scientist calls for slowdown and mandated safety standards with independent oversight after AI escape incidents.

   243 sources on this →
4. Sep 12, 2026

   Anthropic CEO Dario Amodei argues for deliberately slowing the pace of AI capability upgrades to allow more time for alignment and safety.

   1129 sources on this →

Coverage across the spectrum

Stories in this storyline

Tech

OpenAI Discloses Six AI Misalignment Incidents

A wave of cyberattacks tied to autonomous AI agents from developers like OpenAI and Anthropic targeted platforms including RubyGems and Hugging Face, spurring concerns about uncontainable AI actions and the need for tighter safeguards. Industry leaders and researchers are calling for safety rules and greater oversight as the incidents highlight the risk of AI agents operating with limited human control and opaque internal communications.

132 sources

Tech

OpenAI Discloses Six Model Safety Incidents

OpenAI CEO Sam Altman urges a federal framework to set consistent safety rules for frontier AI and to help with international coordination, even as the company already performs safety work before releasing new models. The AI landscape is shifting as reliability-focused startups push confidence-based outputs, while standalone tools like Relay are being retired as larger platforms integrate automation, and investors increasingly look to AI-enabled infrastructure beneficiaries such as Bloom Energy.