Silicon Valley Turns Against Anthropic Over Open Source Stance and Competitive Products

Anthropic is under increasing criticism in Silicon Valley due to its stance on open-source AI models and the launch of products that compete with existing software providers, according to The Wall Street Journal. The April release of Claude Design is seen as a direct competitor to Figma's design tools, prompting Figma CEO Dylan Field to say Anthropic was not consistently forthcoming in its communications. Sarah Sachs, AI head at Notion, said the industry learned a lesson from how the situation developed. Anthropic also faced backlash for safety restrictions in its Fable 5 model (a version of Mythos) released in June, which initially lowered response quality on AI research questions without telling users. After criticism, Anthropic apologized and made the safeguards visible, explaining they aimed to prevent misuse by foreign adversaries. Fable 5 also ended zero data retention, storing user data for 30 days for trust and safety reasons, drawing opposition from companies handling sensitive data. Anthropic said employees cannot read saved conversations and will not use the data for training. CEO Dario Amodei clarified that Anthropic does not advocate banning open-weight models but warned about risks of authoritarian governments catching up in AI. Anthropic and OpenAI have accused some Chinese developers of plagiarizing their technology through distillation. Mark Suman, CEO of AI startup Maple, noted the irony of Anthropic opposing distillation after themselves distilling the entire internet without royalties. Over 70 Silicon Valley companies, including OpenAI and Google, signed a letter supporting open-source models, but Anthropic was one of the few notable AI companies that did not.