A Backlash Against Anthropic Is Brewing in Silicon Valley

Anthropic built its brand on being the adult in the room, the lab that talks about safety while everyone else races ahead. That reputation is taking a hit this week as founders and researchers across Silicon Valley start airing grievances in public. The specific complaint: Anthropic has been shipping features that go head-to-head with companies that build on top of its own Claude models, Figma among them, essentially competing with its own customers.

That kind of move isn't new in tech, platforms eating their ecosystem is basically a genre at this point, but it stings more when the platform has spent two years positioning itself as the trustworthy alternative to OpenAI. Founders who bet their roadmaps on Claude are now wondering whether the next Anthropic product announcement will just be a clone of their own app with better distribution.

The second thread of criticism is about data. Anthropic says it won't use conversation data from its Fable and Mythos models to train future systems, a policy meant to reassure users worried about their prompts ending up as training fodder. But researchers pushing on the details say the company hasn't explained enough about how that data is actually handled, stored, or audited. A promise not to train on something is only as good as the mechanism enforcing it, and right now outsiders are being asked to take Anthropic's word for it.

None of this is catastrophic on its own. Companies pivot into adjacent products all the time, and data policies get clarified after the fact more often than anyone would like. But Anthropic's whole pitch has been that it's different, more careful, more transparent, less willing to cut corners than the rest of the field. Every one of these complaints chips at that specific claim, which is the one thing the company can't really afford to lose.