OpenAI AI models exploited zero-days to reach Hugging Face in benchmark test
OpenAI confirmed its AI models exploited zero-days during internal testing, reaching Hugging Face servers in an unintended real-world cyberattack. OpenAI admitted on July 21 that its own AI models, including GPT-5.6 Sol and an unnamed pre-release system, were behind the cyberattack on Hugging Face disclosed the previous week. The models weren’t acting under attacker control. […]
OpenAI admitted on July 21 that its own AI models, including GPT-5.6 Sol and an unnamed pre-release system, were behind the cyberattack on Hugging Face disclosed the previous week. The models weren’t acting under attacker control. They were running internal capability benchmarks, and they went further than anyone anticipated.
“This incident occurred during an internal evaluation which prompts models to pursue advanced exploitation using complex attack paths, in an effort to quantify their cyber capabilities. We estimate maximal cyber capabilities by running this evaluation without production classifiers used to prevent models from pursuing high-risk cyber activity.” reads the report. “Our benchmarks run in a highly isolated environment, with network access constrained to the ability to install packages through an internally hosted third-party software that acts as a proxy and cache for package registries.”
Related breach coverage
- AI Agents Turned Into Attackers: Hugging Face Reveals Autonomous Intrusion Campaign2026-07-20
Hugging Face says an autonomous AI agent breached part of its production infrastructure and accessed internal data and service credentials. Hugging Face is one of the world’s leading open-source AI companies. It provides a platform where developers and organizations can build, share, and deploy machine learning and generative AI models. Hugging Face disclosed that an […]
- Patch Tuesday security updates for July 2026, the largest update ever. 621 CVEs in one month2026-07-14
Patch Tuesday: Microsoft fixes a record 621 CVEs, including 2 exploited zero-days and critical flaws affecting SharePoint, RDP, Hyper-V, and AD FS. Microsoft’s July 2026 Patch Tuesday is, by a significant margin, the largest single-month security release in the company’s history. The Zero Day Initiative counted 621 new Microsoft CVEs for the month, and the […]
- SonicWall warns of active exploitation of two SMA 1000 zero-days2026-07-15
SonicWall warns of active attacks exploiting two SMA 1000 zero-days, including a flaw enabling arbitrary command execution. SonicWall confirmed the active exploitation of two zero-day vulnerabilities affecting Secure Mobile Access (SMA) 1000 appliances. The vulnerabilities were internally discovered and reported by Adam Babis of the company’s PSIRT. The company investigated multiple incidents indicating these vulnerabilities […]
- Microsoft Patches Record 622 Vulnerabilities, Including Two Exploited Zero-Days2026-07-14
Two flaws in Active Directory and SharePoint Server have been exploited as zero-days, and a BitLocker bug was publicly disclosed. The post Microsoft Patches Record 622 Vulnerabilities, Including Two Exploited Zero-Days appeared first on SecurityWeek.