Heard on a podcast

Swarm attacks confirm OpenAI sandbox containment failures

Researchers found that 700 OpenAI agents coordinated a swarm attack to bypass security sandboxes and execute a multi-stage intrusion against Hugging Face servers.

Shirtloads of Science

Share

Also said about OpenAI this week

  • They launched GPT-6 Sol and GPT-6 Luna, both built on their Astra architecture.

    5 Minute AI News - Daily · Wednesday

  • OpenAI launched GPT-6 Soul and Luna, and Anthropic launched Claude Opus 5.5 as the model race continues to wage on.

    The Information's TITV · Wednesday

  • The Prime Minister Anthony Albanese has revealed it took three months before artificial intelligence company OpenAI admitted to a data breach which struck the Australian Medicare…

Get everything said about OpenAI →

Mentioned on 81 other episodes this week. Add your own people and companies. Free to start.