https://pentforge.com/insightsPlain-language defensive security explanations.https://pentforge.com/insights/2026-10-31-github-s-secret-scanning-uses-ai-to-reduce-false-positiveshttps://pentforge.com/insights/2026-10-31-github-s-secret-scanning-uses-ai-to-reduce-false-positivesSat, 31 Oct 2026 09:00:00 GMTGitHub improved secret detection by combining regex with a context-aware AI classifier, cutting false positives and scan times while scaling to push protection.https://pentforge.com/insights/prompt-injection-is-a-permission-problemhttps://pentforge.com/insights/prompt-injection-is-a-permission-problemSat, 31 Oct 2026 09:00:00 GMTA model can mistake outside content for instructions. The damage depends on what the surrounding application lets it do.https://pentforge.com/insights/a-scan-is-not-a-remediation-programmehttps://pentforge.com/insights/a-scan-is-not-a-remediation-programmeSat, 31 Oct 2026 09:00:00 GMTFinding a weakness, changing a system and proving the change worked are different jobs. Scope them separately.