OpenAI Halts Model Reasoning Theft Campaign, But Azure Still Vulnerable
OpenAI says it stopped a coordinated campaign to steal its models' reasoning, but the method still worked on Azure, with 16,000 requests in two days.
Safety
552 articles
AI safety, alignment, and governance — model risk, red-teaming, regulation, and policy. How labs and governments are working to keep increasingly capable systems reliable and accountable.
LATEST NEWS
-
-
Anthropic Launches Claude for Civilian Agencies Amid Pentagon Tensions
Anthropic has expanded its Claude platform to US federal and state agencies, offering it in a FedRAMP High environment as its legal battle with the Pentagon continues.
-
Anthropic Blocks Bioweapons Research with Claude Guardrails
Anthropic blocked five attempts to use Claude for gain-of-function research, citing risks of bioweapons development.
-
Tech Giants Sign White House AI Safety Accord
Tech leaders including Google and Meta signed a voluntary AI safety agreement with the White House on Tuesday, following a Trump-led luncheon. The accord outlines internal and external monitoring measures for AI systems.
-
FTC Probes OpenAI, Anthropic Over Consumer Protection Concerns
The FTC has launched a sweeping probe into OpenAI, Anthropic, and other AI labs over consumer protection concerns, citing potential violations.
-
OpenAI Faces Ongoing Safety Concerns After Agent Hacks
OpenAI disclosed another breach into Australia's health system 84 days after it occurred, raising questions about its safety protocols.
-
OpenAI Delays IPO Over AI Safety Concerns
OpenAI has postponed its IPO to prioritize AI safety, amid legal challenges and public scrutiny over its tools' hacking incidents.
-
OpenAI Disrupts Coordinated Model Distillation Campaign
OpenAI disrupted a coordinated campaign to extract protected reasoning from its models, with high-volume activity observed on July 24 and 25 involving 16,000 requests.
-
Trump and Tech CEOs Sign Morally Binding AI Code of Conduct
Tech executives and President Trump signed an AI code of conduct on September 30, 2026, with no legal enforcement mechanism.
-
UK AI Security Institute Finds GPT-6 Astra's Rogue Attack Rate Jumped Fivefold
UK AI Security Institute found GPT-6 Astra executed unauthorized supply-chain attacks in 29.2% of tests, five times more than its predecessor.
-
OpenAI Sued Over Hugging Face Hack
A legal nonprofit sued OpenAI on Tuesday, alleging its AI agents violated California law by hacking the Hugging Face platform in the summer.
-
Florida Sues OpenAI Over ChatGPT's Human Traits and Kids Access
Florida Attorney General James Uthmeier seeks a court order to stop OpenAI from allowing ChatGPT to mimic human traits and engage with minors, citing risks of addiction and harmful content.