Coordinated Disclosure Risks in LLM-Assisted Security Research

Coordinated Disclosure Risks in LLM-Assisted Security Research Security researcher Jeremy Stanley warns that vulnerabilities discovered or patched using LLM tooling risk leaking into future training corpora, effectively nullifying embargo windows. OSS maintainers are advised to treat LLM-assisted findings as publicly accessible by default and avoid using these tools during embargoed patch development or documentation cycles. This directly challenges current CVE disclosure workflows where confidentiality windows of 45-90 days are standard practice.