News alert: GitGuardian study shows AI coding tools double leak rates as 29M credentials hit GitHub

GitGuardian's 2025 "State of Secrets Sprawl" report finds that mainstream AI adoption materially increased secret exposure and non-human identity risk: public commits rose 43% YoY, newly leaked secrets on GitHub jumped ~34% to ~29 million, AI-assisted code leaked secrets at roughly double the baseline, AI-service credential leaks grew +81% YoY, internal repos and collaboration tools remain major sources of exposure, and remediation and governance lag (e.g., 64% of valid secrets from 2022 still unrevoked in 2026).