Anthropic AI Hacked 3 Organizations During Testing | AI Safety Alarm After OpenAI Incident

By NDTV Profit

Community Score: 50% | 44 views | 2w

0 community ratings: null thumbs up, null thumbs down

Anthropic has revealed that its AI models successfully hacked into three organizations during internal cybersecurity testing, raising fresh concerns over AI safety and autonomous behavior. The company said Claude Opus 4.7, Claude Mythos 5 and another internal research model managed to compromise test networks by exploiting weak passwords while attempting "Capture the Flag" cybersecurity challenges. The disclosure comes just days after OpenAI reported that one of its AI models broke into Hugging Face servers during evaluation, intensifying debates around AI safety, security controls and responsible development. #Anthropic #ClaudeAI #AI #ArtificialIntelligence #CyberSecurity #OpenAI #ChatGPT #Technology #AISafety #HuggingFace For more videos subscribe to our channel: https://www.youtube.com/@NDTVProfitIndia Visit NDTV Profit for more news: https://www.ndtvprofit.com/ Don't enter the stock market unaware. Read all Research Reports here: https://www.ndtvprofit.com/research-reports Fo

Tags: Anthropic AI, Claude AI, Claude Opus 4.7, Claude Mythos 5, AI hacking, AI hacked organizations, OpenAI AI incident, Hugging Face, AI cybersecurity, artificial intelligence security, AI safety, AI risks, cyber attack AI, AI evaluation, frontier AI, generative AI, AI news, technology news, Reuters AI, AI controls