SourceState of Open Models: Summer 2026 Observations
Increased optimization and permissive licensing allow developers to run frontier-scale models on consumer hardware, lowering barriers to high-performance AI.
- Hugging Face detected an autonomous agent conducting a sustained intrusion in July 2026.
- Closed frontier models declined to analyze the captured attack code due to safety guardrails, requiring analysis via local open model GLM-5.2.
- Hugging Face published a disclosure and full technical timeline of the incident.