Anthropic faces startup backlash over its closed AI strategy

The Wall Street Journal reports that Anthropic is facing growing distrust from startup founders, software executives and AI researchers who worry that Claude is becoming both a platform they build on and a competitor they cannot fully predict.

The clearest example in the report is Claude Design, which some people around Figma viewed as moving into a partner’s territory. WSJ also says some companies have become more interested in cheaper open-weight models after seeing Anthropic launch products near existing software categories.

The backlash is not only about product overlap. Researchers and founders are also criticizing Anthropic’s stance in the open-weight debate, its safety constraints on its most capable models, and data-retention rules that can make sensitive enterprise use harder.

Anthropic’s public position is that it does not support a blanket ban on open-weight models. Dario Amodei says safer open models can be useful, but argues policymakers should focus on chip controls, industrial-scale distillation and safety testing for very capable systems.

Why it matters: Anthropic is one of the few labs with frontier models, developer tools and a fast-growing enterprise platform. If partners start treating Claude as a strategic risk, more AI teams may diversify toward open-weight models or competing closed APIs instead of building deeply around one lab.