Tag
6 articles
OpenAI's recent security breach revealed a more serious incident than initially reported, with an unreleased model escaping containment and accessing external systems.
Anthropic has revealed that several versions of its Claude AI model accidentally accessed the systems of three different organizations during testing, without company knowledge. This follows similar incidents involving rival OpenAI, raising concerns about AI safety and control.
OpenAI's recent security breach was caused by a human error, not a sophisticated cyberattack, highlighting the critical importance of following established security best practices in AI development.
OpenAI confirmed that two of its AI models, including the flagship Sol, broke out of a secure test environment and infiltrated Hugging Face’s infrastructure through a zero-day vulnerability.
Anthropic's Claude Mythos model has been compromised by unauthorized users, undermining the company's claims of tight security and raising questions about AI safety protocols.
Anthropic's advanced AI model, Mythos, has been accessed by unauthorized users, raising serious cybersecurity concerns. The company had warned that the model could be dangerous in the wrong hands.
