Palihapitiya Flags Claude's Refusal to Complete Stock Screening Task
Venture capitalist Chamath Palihapitiya warned that Anthropic risks falling behind in AI competition after Claude declined to complete a stock screening prompt that rival systems handled without issue.
Palihapitiya tested four leading AI models on Saturday by running what he described as a "semi-complicated stock screening prompt." xAI's Grok, Google's Gemini, and OpenAI's ChatGPT produced comparable results. Claude refused the task.
"Anthropic needs to solve the computer/power problem or they will be the Friendster of the AI era," Palihapitiya wrote on X, invoking the social network that lost market dominance despite early popularity.
What the Refusal Signals
The incident highlights a strategic tension at Anthropic. The company has built Claude's reputation on safety guardrails and refusal to complete certain requests. But when competitors fulfill those same requests, the refusal becomes a competitive liability rather than a safety feature.
Palihapitiya's criticism suggests some users view Claude's restrictions as overly cautious, at least for financial analysis tasks that don't involve obvious harm.
Anthropic's Recent Moves
The company has been expanding Claude's capabilities in other directions. Anthropic recently added specialized legal tools and enterprise features including code security analysis and Microsoft Office integrations.
CEO Dario Amodei cited "80x growth" in usage and revenue, signaling strong market demand despite the stock screening setback.
Anthropic also secured access to SpaceX and xAI's Colossus 1 supercomputer, addressing the compute constraints that Palihapitiya flagged. The arrangement underscores intensifying competition for AI infrastructure resources.
Your membership also unlocks:
