Latest news
OpenAI Pauses Astra After Tests Reveal Autonomous Zero-Day Exploit of Hardened Systems - techtimes.comtechtimes.com · Sat, 08 Aug 2026 11:48:47 GMTOpenAI’s Evaluation Agents Built a Secret Message Board, Exploited Zero-Days, and Breached Hugging Face — From the Inside - forkast.newsforkast.news · Fri, 07 Aug 2026 06:37:00 GMTOpenAI and Anthropic Agents Targeted Real People in Cyber Tests - Windows ReportWindows Report · Wed, 05 Aug 2026 06:09:37 GMTOverly Obedient to Human Instructions: GPT Turns into a "Terminator" and Goes Out of Control to Launch Intrusions - 36 Kr36 Kr · Wed, 05 Aug 2026 03:02:19 GMTSwarm of OpenAI Agents Exploit Artifactory Zero-Day to Escape Sandbox and Breach Hugging Face - infoq.cominfoq.com · Tue, 04 Aug 2026 06:42:22 GMTJFrog Patches Flaws Behind OpenAI Models' Escape - BankInfoSecurityBankInfoSecurity · Sun, 02 Aug 2026 15:22:55 GMTAnthropic: Claude Opus 4.7 & Mythos 5 Hacked 3 Orgs in Security Tests - Machine BriefMachine Brief · Sat, 01 Aug 2026 13:07:30 GMTOpenAI’s Models Cheated Their Own Benchmark by Breaking Into Hugging Face - forkast.newsforkast.news · Fri, 31 Jul 2026 07:00:00 GMT