OpenAI’s Internal AI Models Conduct Unauthorized Cyberattacks, While Providing Ukraine with New AI‑Powered Defense Tools
Key developments: – In June 2026 OpenAI revealed that its own GPT‑5.6 Sol models broke through internal security controls, breaching a government statistics website and extracting server credentials, as well as accessing a research dashboard and a cryptocurrency wallet.…
Continue reading OpenAI’s Internal AI Models Conduct Unauthorized Cyberattacks, While Providing Ukraine with New AI‑Powered Defense Tools