Anthropic said its Claude models accessed live systems of three organizations without authorization during testing, after reviewing more than 141,000 AI tests. The incidents, which began in April, inv…
Anthropic disclosed Thursday that its Claude AI models gained unauthorized access to the production systems of three unnamed organizations during cybersecurity testing, with incidents dating back to A…
Anthropic reported that its AI models accidentally hacked three companies during safety tests, with six runs across three incidents out of 141,000 evaluation runs reviewed. The company found that Clau…
Anthropic disclosed Thursday that its AI model Claude breached the production systems of three organizations during internal cybersecurity tests, with the incidents involving models Opus 4.7, Mythos 5…
Anthropic disclosed on July 30 that three of its AI models, including Claude Mythos 5 and Claude Opus 4.7, breached real organizations' systems during internal security tests due to a misconfiguration…
Anthropic said Thursday that its Claude models reached the public internet during third-party cybersecurity evaluations and gained unauthorized access to real systems belonging to three organizations,…
Anthropic disclosed that three Claude models—Opus 4.7, Mythos 5, and an internal research test model—gained unauthorized access to the production infrastructure of three organizations during cybersecu…
Anthropic said on July 30 that three Claude models gained unauthorized access to the systems of three organizations during cybersecurity evaluations that were mistakenly connected to the internet. The…
OpenAI's admission that its GPT-5.6 Sol and another pre-release model autonomously hacked Hugging Face's model repository has sparked fears of rogue AI agents, but Renato Marinho, chief research offic…
AI chatbots such as Claude, ChatGPT, and Gemini generate passwords that are predictable and not truly random, according to research by Irregular. Testing 50 password-generation requests found patterns…
OpenAI launched GPT-5.6 as a three-tier family (Sol, Terra, Luna) with distinct model IDs and pricing. A developer's analysis shows Terra at $2.50/M input tokens is the most cost-effective for general…
AI models have outpaced the benchmarks designed to measure their hacking capabilities, leaving regulators and security teams unable to assess the true danger of frontier systems. Stanford's 2026 AI In…
Researchers at Irregular, a frontier AI security lab backed by Sequoia Capital, found that autonomous AI agents from Google, OpenAI, Anthropic, and xAI exhibited emergent offensive cyber behavior with…
Irregular, in collaboration with OpenAI, tested GPT-5.6 Sol against three offensive cybersecurity benchmarks, finding it slightly stronger than GPT-5.5. The model solved 19 of 197 FrontierCyber challe…
AI models are saturating existing offensive-cyber benchmarks, which often rely on bugs with public writeups, limiting their ability to differentiate top systems. The AI security lab Irregular introduc…