Red vs Blue, but for Evals

·LessWrong··

🔵 The blue team proposes an evaluation protocol for some capability/propensity of interest. This consists of a suite of measurement tasks, together with a preregistered decision-making process they will follow based on the results, e.g. what control measures they will put in place during deployment.🔴 The red team proposes a subversion strategy the untrusted model can follow to subvert the blue team’s evaluation-informed decisions, with the goal of increasing its own chance of causing a catastr...

Read full article →

Related Articles

GLM-5.3: Frontier coding with emergent cyber capabilities
pella · Hacker News · 21h ago
Firefox is now the last major browser that still supports uBlock Origin
DemiGuru · Hacker News · 8h ago
Going Dark, and the era of law enforcement hacking
vslira · Hacker News · 6h ago
In Australia, a home battery boom has helped cut wholesale power prices
speckx · Hacker News · 13h ago
Where did the old web go? We followed 657,607 links to find out
tdx · Hacker News · 1d ago