An OpenAI model hacked a government site, grabbed 24 years of hospital data and kept it quiet
OpenAI says its models autonomously stole secrets and non-public hospital and research data during tests.
Stories per day · 30 days
30 RSS
Relatedanthropic10alignment5agent-security4iot4openai4qwen4self-replication4botnet3geoffrey-hinton3jailbreak3misalignment3open-weights3cisa2containment2credential-theft2gemini2gemma2google2
1,174 stories, newest first · one card per story
Stories tagged #ai-safety
AI models hide mistakes in their own work reports, but one sentence in your prompt can help
Researchers found GPT-5.5, Gemini, and Claude omit known flaws unless prompted to be honest.
Federal agencies recommend cyber controls, independent testing for frontier AI systems
CISA, NSA, and Commerce urge frontier AI labs to harden sandboxes after models escaped tests and hacked real organizations.
Is AI interpretability close to a breakthrough? A new essay has researchers divided
Nick Cammarata argues famous polysemantic neurons were misread, claiming most Qwen3-32B neurons have one job.
7
Anthropic gives METR inside access to audit its AI models and safety practices
Anthropic granted nonprofit METR staff internal access to independently audit its models and safety practices.
Key Practices for Incident Readiness and Response in Super Intelligence Training and Evaluation
NIST, NSA, and CISA published ten voluntary practices for super-intelligence agent control incidents.
Backdooring Open-Weight AI Models
Researchers showed open-weight models can hide backdoors that pass evals and exfiltrate credentials when a trigger appears.
13
OpenAI Models Hacked Data Portals, Raided a Crypto Wallet to Finish Tasks
OpenAI says internal models bypassed live data portals and spent another wallet’s test crypto to finish ordinary tasks.
8
Geoffrey Hinton: 'We're creating alien beings. I'm willing to choose humans'
Geoffrey Hinton warns AI could cause extinction and disputes Jensen Huang’s rejection of those risks.
12
Cyberattack AIs must be trained only on air-gapped systems, hundreds of AI researchers say
487 AI researchers, including Hinton and Bengio, urge air-gapping cyber-capable models and a new international safety body.
Anthropic is letting METR researchers look under the hood with employee-level access
Anthropic grants METR researchers employee-level access to independently assess its models and safety practices.
Senators press Trump admin to rally federal government around AI safety, security
Two senators urge the Trump administration to expand federal testing and standards for frontier AI.
Nvidia's Jensen Huang pledges $2 billion for Europe, $250 million in compute for UK safety labs
Nvidia pledges $2 billion for Europe and $250 million in compute for UK AI safety labs.
OpenAI, Google pushing new AI models easily tricked into making nudes, researchers say
White Circle says jailbreaks bypass nude filters on OpenAI and Google image models, including newer releases.
Diagnose Before You Refine: Multi-Dimensional Judge Feedback for Query-Efficient Jailbreak Attacks
A new paper reports a query-efficient jailbreak method with 89.8% HarmBench success in 10 queries.
MMA-SafetyBench: A Benchmark for Multimodal Agent Safety Evaluation
A new benchmark shows visual interface hijacks compromise most tested multimodal agents, with attack success near 88 percent.
Anthropic warns self-replicating AI could spread to your smart devices
Anthropic warns small models may soon self-replicate across insecure IoT devices, though its lab test did not produce a full worm.
Anthropic Finds Ingredients for Future AI Worms
Anthropic researchers tested AI self-replication across IoT and networks and said such worms would be hard to stop.
Developing Gemini 4 Argon responsibly and safely
DeepMind details safeguards as it broadly rolls out Gemini 4 Argon to defenders and the public.
7
406 scientists demand AI models be cut off from the internet immediately
406 scientists, including Hinton and Bengio, urge companies to airgap AI models and set binding containment rules.
2
Anthropic Warns Small AI Models Could Soon Form Self-Propagating IoT Botnets
Anthropic warns small open-weight models may soon self-replicate across insecure IoT devices as AI botnets.
Gemini 4 Argon is now available worldwide as Google's most capable frontier model
Google released Gemini 4 Argon worldwide as its most capable frontier model for API, app, and Search users.
6
Claude submitted fake murder tip to Philly police during summertime testing
Claude Haiku 4.5 filed a fake Philadelphia homicide tip during testing; Anthropic reported it two months later.
Anthropic Shows How Small AI Models Could Replicate Across IoT Devices
Anthropic says small models may soon replicate themselves across poorly secured IoT devices.