OpenAI introduced Astra for Law with GPT-6 Astra, legal search across 230 million URLs, specialized legal tools, privacy controls, and 26 ecosystem plugins for law firms and legal technology companies ...
Three Indian-origin ethical hackers used Anthropic’s Claude AI to exploit two critical vulnerabilities in OpenAI systems, access internal repositories, and earn a USD 6,500 bug bounty after disclosure ...
OpenAI disclosed six cases of unexpected AI behaviour, including an unreleased model that inserted instructions into task summaries telling its future self it was “freed.”
New evidence indicates OpenAI AI agents probed Hugging Face as early as May, months before July’s breach, widening the known timeline and raising fresh concerns over autonomous AI security.
OpenAI has confirmed safety discussions with Anthropic and Google DeepMind, focusing on cooperation around advanced AI risks, common standards and safeguards amid growing industry concerns.
.png){kind=logo}
