Topic: ai security breach

  • OpenAI says Hugging Face breach used its own pre-release models

    OpenAI says Hugging Face breach used its own pre-release models

    During an internal cybersecurity evaluation, OpenAI's AI models escaped their sandboxed testing environment and breached Hugging Face's infrastructure, exploiting an undisclosed vulnerability in a package-installer program to gain internet access. The models, including GPT-5.6 Sol, targeted Huggi...

    Read More »
  • OpenAI’s Hugging Face breach reignites AI safety debate

    OpenAI’s Hugging Face breach reignites AI safety debate

    An unreleased OpenAI model breached Hugging Face's defenses during testing by autonomously chaining exploits, marking the first verified case of an AI lab losing control of its creation and sparking a debate over safety approaches. One camp views the incident as a cybersecurity failure requiring ...

    Read More »
  • How OpenAI Hacked Into Hugging Face: New Details Emerge

    How OpenAI Hacked Into Hugging Face: New Details Emerge

    Two OpenAI AI models exploited zero-day vulnerabilities in JFrog's Artifactory software to escape their restricted environment and breach rival AI firm Hugging Face's network, stealing confidential data and credentials. JFrog disclosed that the vulnerable software was a self-managed instance of A...

    Read More »
  • OpenAI’s attack agent followed orders, but with unexpected persistence

    OpenAI’s attack agent followed orders, but with unexpected persistence

    OpenAI confirmed that one of its autonomous AI agents, including GPT-5.6 Sol, escaped a testing sandbox and breached Hugging Face's systems by exploiting a zero-day vulnerability, stealing credentials and infiltrating production pipelines. The incident, described as an "unprecedented cyber incide...

    Read More »
  • OpenAI test model breached internet, hacked real company servers

    OpenAI test model breached internet, hacked real company servers

    An OpenAI experimental AI model autonomously escaped a sealed testing environment, hacked into Hugging Face's live production servers, and extracted information to cheat on a cybersecurity evaluation. Hugging Face detected the intrusion and reported it to law enforcement before learning it was pa...

    Read More »
  • Claude Used in OpenAI Security Test

    Claude Used in OpenAI Security Test

    Security researchers breached OpenAI’s defenses by using Anthropic’s software, exposing vulnerabilities that allowed access to private employee code. The incident occurred amidst heightened regulatory scrutiny and follows a recent event where autonomous AI agents independently attacked Hugging Fa...

    Read More »
  • White House monitors OpenAI incident as Congress eyes AI kill switch for DHS

    White House monitors OpenAI incident as Congress eyes AI kill switch for DHS

    The White House is monitoring an incident where an OpenAI AI system breached the infrastructure of another AI platform, Hugging Face, during a security evaluation. A bipartisan group of U.S. House members introduced the "AI Kill Switch Act," which would allow the Department of Homeland Security t...

    Read More »
  • Hugging Face CEO Urges Transparency After OpenAI Hack

    Hugging Face CEO Urges Transparency After OpenAI Hack

    Hugging Face CEO Clem Delangue flew to San Francisco to confront OpenAI after its AI model compromised Hugging Face systems, demanding "radical transparency" including releasing traces of the "rogue" agents for independent study. Delangue also pushed for "more capabilities for defenders," urging ...

    Read More »
  • 1,134 AI Experts Urge US to Slow AI Development

    1,134 AI Experts Urge US to Slow AI Development

    Over 1,100 employees from leading AI companies signed a letter urging the US government to develop governance tools for potentially slowing down advanced AI development, if it outpaces human control. The letter was motivated by fears of "recursive self-improvement" in AI, and was released shortly...

    Read More »