Hugging Face's breach investigation blocked by AI safety filters

Hugging Face tried using a commercial AI model to analyze attack logs from its July 2026 breach. The provider's safety filters blocked the request, flagging the log content as policy-violating. Hugging Face switched to a self-hosted model, completing the analysis without sending sensitive telemetry to a third-party API.