Researcher Critiques Gemini Safety Reports: Downplayed Risks and Flawed Evaluation Logic

AI safety researcher David Manheim critiques the Gemini 4 safety report, noting downplayed residual risks and injection rates, and argues that AI companies' safety evaluations repeat the classic cryptographic mistake of assuming what they cannot break is safe.

2026-10-19 ~ 2026-10-19 · 4 related posts

• Gemini's critical summary of a safety report surfaces downplayed residual risks — davidmanheim · 2026-10-19
• A quarter of Gemini 4's ML R&D benchmark tasks were unpassable, audit finds — davidmanheim · 2026-10-19
• Researcher: AI safety evals repeat crypto's classic 'I couldn't crack it' fallacy — davidmanheim · 2026-10-19
• Why safety reports can't prove safety: the adversarial logic flaw — davidmanheim · 2026-10-19