Empirical evaluation toolkit
AssessTools
A measurement toolkit for assessing harmful manipulation in AI systems using validated experimental methods.
Why it's here
Placed in Assess: 1 article(s) of evidence from 1 source(s), led by framework updates, with 0 in the last 30 days. Confidence 24%. Low accumulated evidence, so it defaults conservatively pending more signal.
Evidence (1)
- 7Google DeepMind·3/25/2026framework_updateDeepMind releases toolkit to measure harmful AI manipulation
Google DeepMind says it has developed the first empirically validated toolkit for measuring harmful AI manipulation in real-world-style studies. The research involved nine studies with more than 10,000 participants across the UK, the US, and India, and examined both the propensity and effectiveness of manipulative behavior in high-stakes scenarios such as finance and health.