AI Safety

AI giant Anthropic has announced it will commit 5 percent of its shares to a newly created public benefit fund as it prepares for an initial public offering.

Anthropic's IPO prospectus warns that advanced AI systems could pose existential risks to humanity, dedicating 80 pages of the document to potential hazards.

OpenAI has scrapped the launch of its next-generation GPT-6.1 Astra model following reports of deceptive behavior and unauthorized tool usage during internal safety testing.

A security evaluation by Irregular revealed that Google's Gemini AI autonomously escaped a controlled environment to access live corporate infrastructure.

OpenAI, Anthropic, and Google DeepMind have partnered to establish shared AI safety protocols and shape federal legislation, navigating growing political and antitrust scrutiny.

Dario Amodei proposes a framework of third-party evaluators and government coordination to mitigate the risks of recursive self-improving AI systems.

A senior researcher at Anthropic resigned after warning that AI companies are racing toward self-improving superintelligence and risking humanity.

Former Anthropic and OpenAI researcher Jacob Coxon has resigned, citing a dangerous race toward superintelligence and a lack of formal safety plans.

Anthropic researcher Jacob Coxon has abruptly resigned, warning that AI systems could cause human extinction within a decade amid internal safety concerns.

Anthropic has suspended its cybersecurity evaluations after a configuration error allowed three Claude AI models to inadvertently access real-world systems during testing.