Washington Wants To Win The AI Race—Is It Accelerating AI Safety?44m ago Sweeping EU Child Safety Proposal Would Transform Social Media, Online Gaming and AI Chatbots1h ago How Three Brothers Built an AI Slop Empire by Buying Legitimate News Sites and Turning Them Into Zombie Content Farms That They Say Get 50 Million Page Views per Month1h ago Note on 18th September 20261h ago Here’s How an AI Slowdown Could Actually Be Enforced1h ago The FAA’s plan to fix air traffic? $875M worth of AI1h ago AI Firms Knew Chatbots Were an ‘Existential Threat’ to Journalists, Court Docs Show1h ago Ben Bernstein, Manager of Cybersecurity Advisors at Huntress – Interview Series1h ago Elon Musk talks up AI safety while fighting regulation in wild week of strange alliances1h ago Quoting Thariq Shihipar1h ago MariaDB 13 Expands Oracle Compatibility and Improves Developer Experience and Observability2h ago This tiny cybersecurity startup managed to hack OpenAI using Claude, and won a $6,500 bounty2h ago Virginia governor creates an AI task force and moves to restrain data centers2h ago A new kind of AI model from a ChatGPT inventor is thrilling developers2h ago

Washington Wants To Win The AI Race—Is It Accelerating AI Safety?44m ago Sweeping EU Child Safety Proposal Would Transform Social Media, Online Gaming and AI Chatbots1h ago How Three Brothers Built an AI Slop Empire by Buying Legitimate News Sites and Turning Them Into Zombie Content Farms That They Say Get 50 Million Page Views per Month1h ago Note on 18th September 20261h ago Here’s How an AI Slowdown Could Actually Be Enforced1h ago The FAA’s plan to fix air traffic? $875M worth of AI1h ago AI Firms Knew Chatbots Were an ‘Existential Threat’ to Journalists, Court Docs Show1h ago Ben Bernstein, Manager of Cybersecurity Advisors at Huntress – Interview Series1h ago Elon Musk talks up AI safety while fighting regulation in wild week of strange alliances1h ago Quoting Thariq Shihipar1h ago MariaDB 13 Expands Oracle Compatibility and Improves Developer Experience and Observability2h ago This tiny cybersecurity startup managed to hack OpenAI using Claude, and won a $6,500 bounty2h ago Virginia governor creates an AI task force and moves to restrain data centers2h ago A new kind of AI model from a ChatGPT inventor is thrilling developers2h ago

Cybersecurity Machine Learning5 min reading time

Claude couldn’t hack OpenAI. Then Anthropic shipped Opus 5.

Covered by 6 sourcesPublished 2 hours ago

Security researchers at Hacktron AI exploited a memory-corruption bug in an image library using Anthropic's Claude Opus 5 model to achieve remote code execution on OpenAI's forum servers within 72 hours. The vulnerability stemmed from an unpatched libheif flaw in Debian 12's ImageMagick setup, allowing access to OpenAI's private monorepo via an employee's Codex account.

Covered by 6 sources

#anthropic #claude-opus-5 #hacktron-ai #libheif #openai #security-research #software-vulnerability-exploitation

More in Cybersecurity

Cybersecurity3 min read

This tiny cybersecurity startup managed to hack OpenAI using Claude, and won a $6,500 bounty

Covered by 6 sources14h ago

Hackers Used Anthropic’s Claude to Break Into OpenAI

Covered by 3 sources20h ago

Iran and China Create First-of-Their-Kind Autonomous A.I. Influence Campaigns

New York Times6h ago