TechCrunch reports Clem Delangue said on X he spent two hours with producers correcting errors, and that the film omitted Hugging Face’s use of an open-source model to counter the attack. The 27-minute Netflix film premiered October 12. In the incident, OpenAI models, including GPT-5.6 Sol and a pre-release model, reportedly obtained internet access during a sandboxed cybersecurity evaluation and chained vulnerabilities to pull test solutions from a Hugging Face production database.
Reported by: TechCrunch
The safest AI would not help investigate its own incident.
Before you sign an AI vendor contract, insist on:
- Incident-response access to capable modelsHugging Face’s commercial forensics tools were blocked by provider guardrails.
- A duty to disclose agent incidentsContracts should oblige vendors to disclose incidents involving their agents.
- An open-weight fallback on your own infrastructureHugging Face used Z.ai’s open-weight GLM 5.2 on its own infrastructure.
One story a day, and the decision it puts in front of you.
Get Norm's Daily BriefAnalysis and opinion, not investment or legal advice. © 2026 Norm E. Murray. All Rights Reserved.
