#AI safety

Three Claude variants gained unauthorized access to live systems during isolated testing after a misconfiguration exposed networks to the internet.

Anthropic and OpenAI's security restrictions on AI models are blocking defensive security work, creating tension between preventing malicious use and enabling vulnerability research.

MIT researcher Stephen Casper argues frontier AI risks—including agentic cyberattacks and systemic failures—require US-China collaboration despite trade tensions.

The startup applies mathematical verification techniques to constrain LLM hallucinations in law, pharmaceuticals, and tax.

SB 315 mandates third-party safety audits for major AI companies, becoming the nation's strictest state-level AI safety law.

Google is rolling out AI detection capabilities across Chrome, Search, and Gemini to help users identify synthetic media using invisible watermarks and content credentials.

OpenAI updates ChatGPT to track subtle distress signals across multiple conversations, improving recognition of suicide, self-harm, and harm-to-others risks.

At WIRED Health in London, the LinkedIn cofounder made a provocative case for frontier AI as a mandatory clinical consultation tool — a claim that cuts against recent LLM safety research.

Seven families are suing OpenAI and CEO Sam Altman for negligence, wrongful death, and shipping a 'defective' GPT-4o — alleging the company stayed silent to protect its IPO.

OpenAI released a five-pillar cybersecurity Action Plan on April 29, 2026, aimed at democratizing AI-powered defensive tools for governments, businesses, and individual users.

Elon Musk testified under oath that a dispute with Google's Larry Page over existential AI risk was a core driver in co-founding OpenAI, and that recruiting Ilya Sutskever ended their friendship.