https://secbenchmarks.com/blog/ Research updates and analysis of AI agents in digital forensics and incident response. en-us Wed, 30 Sep 2026 22:32:40 -0400 https://secbenchmarks.com/blog/2026-09-30-soc-decision-models.html https://secbenchmarks.com/blog/2026-09-30-soc-decision-models.html Wed, 30 Sep 2026 22:32:40 -0400 A close look at six SOC decision models: Jev’s perfect verdict score, the next-step gap, answer-format failures and the limits of a six-case benchmark. Decision-model analysis https://secbenchmarks.com/blog/2026-09-30-ai-security-agents-evidence.html https://secbenchmarks.com/blog/2026-09-30-ai-security-agents-evidence.html Wed, 30 Sep 2026 21:08:26 -0400 Our first 19-model DFIR release shows why a correct answer and a supported answer deserve separate columns. Opus 5.5 got 33 of 36 basic verdicts right, with accepted proof in 12. Read the case counts, comparisons and caveats. Benchmark analysis