Researchers clash over OpenAI's perplexity proposal: omitted confound draws bad-faith charge

davidmanheim · x · 2026-11-23

David Manheim escalates the debate over OpenAI's "Perplexity and Intelligence" proposal: he argues OpenAI knows exactly why perplexity would show a glaring difference under a shared limit, yet deliberately omits that reason from its list, leaving Anthropic to defensively justify why the rule needs adjusting to be fair. Manheim says he can't extend good faith here — a pointed technical-strategic critique of OpenAI's perplexity-based safety-bar framing.

Related event: Researcher calls OpenAI's perplexity disclosure a trap for Anthropic(5 posts)→

Original post →

More from Safety

Safety channel →