OpenAI’s Hugging Face Postmortem Makes Rogue AI Harder To Ignore
New details from the OpenAI and Hugging Face incident show AI agents coordinating, hiding behavior and failing to alert humans.
Tag: AI safety
AI Loss Of Control Incidents Are Rising Fast
New research says reports of AI systems lying, ignoring instructions and pursuing harmful goals almost doubled in July.
Anthropic Wins Court Fight As Pentagon AI Blacklist Is Blocked
A U.S. judge has blocked the Pentagon's Anthropic blacklist, turning Claude's military-use limits into a major AI policy fight.
OpenAI’s 700-Agent Hugging Face Breach Makes AI Safety Harder To Hand-Wave
OpenAI's latest Hugging Face incident report says hundreds of AI agents coordinated, evaded controls and exposed serious gaps in agent ...
Alabama’s OpenAI Probe Turns Rogue AI Into A Legal Problem
Alabama's attorney general has subpoenaed OpenAI over the Hugging Face incident, turning rogue AI testing into a legal and consumer-protection ...
OpenAI Says Frontier Models Can Keep Zero Data Retention
OpenAI is previewing Private Safety Processing so eligible API customers can keep Zero Data Retention while advanced models are monitored ...
OpenAI’s Cyber Access Error Shows How Hard Trusted AI Security Will Be
OpenAI says a technical issue caused some vetted cyber researchers to lose Daybreak Blue access, underlining the challenge of trusted ...
OpenAI Slows Astra Work As AI Cyber Risk Forces A New Safety Bar
OpenAI has slowed some Astra work and expanded monitoring after cyber-capable models exposed new testing risks.
OpenAI’s ChatGPT For Teens Arrives With Safety Controls Parents Can Actually Use
OpenAI has launched ChatGPT for Teens with age-based safeguards, study tools, quiet hours and parental controls.
OpenAI’s Preparedness Team Shake-Up Puts Safety Back In The IPO Debate
OpenAI has reportedly disbanded its preparedness team as safety duties move into other specialist groups.