GitHub improves secret scanning with ModernBERT
What happened
On 14 October 2026, GitHub published a new approach to secret detection on its official blog: a two-stage pipeline that replaces its previous LLM-based flow. Regular expressions first generate candidates, then a fine-tuned ModernBERT judges whether they are real credentials using file paths, variable names and context. GitHub says that compared with the old LLM pipeline, user acceptance rose 11.5%, median execution time for post-push scanning fell 86%, service costs dropped by more than 98%, and generic secret detection could be extended to push protection.
Written by AI from the reports · updated 3 d ago
Timeline
Follow the reports to see the event from each side.
- GitHub Blog · AI & MLGitHub’s model for secret detection: Finding more secrets with less noise
GitHub’s context-aware secret classifier improves detection quality, raises weighted acceptance by 11.5%, and reduces median scan execution time by 86%.
Heat of this event
Current heat 3·Peak in the comparable range 9（Oct 14 05:00）·Change over 24 hours in the comparable range -49%
The trend compares only the same accounts observed without a break, so its range can be narrower than the current heat count. Move the pointer or tap the chart to see each hour; with a keyboard, use the left and right arrow keys.