Anthropic: Controversial Practices and Rising Tensions

Anthropic, a major player in the artificial intelligence industry, is at the center of a controversy due to its business practices. This week, the company launched Claude Fable 5, a version of its Mythos model, but limited its performance for users looking to develop competing AI models. This decision sparked a strong reaction from customers, forcing Anthropic to slightly soften its stance by promising to notify clients when less powerful versions would be used.

At the same time, Anthropic recruited partners like Figma, only to launch an AI design tool that directly competes with them. This development led Figma to withdraw from the partnership, and Mike Krieger, Anthropic's product director, left Figma's board. Dylan Field, CEO of Figma, stated at a Sequoia Capital event that Anthropic was "not always forthcoming in its communications." With revenues nearing $50 billion, Anthropic finds itself in direct competition with its own clients, a scenario reminiscent of the platform tactics that led to antitrust cases against Microsoft and Google.

Anthropic's Mythos model is also at the heart of the tensions. The company has restricted some of its features while developing competing applications. This strategy has caused discontent among many clients, partners, and investors. Martin Casado, general partner at Andreessen Horowitz, expressed concerns about restricted access to the most powerful models, predicting that only model creators would have full access. Casado is also an investor and board member of the coding startup Cursor, which is both a client of Anthropic and a competitor to Claude Code.

A year earlier, Anthropic had already launched Claude Code, a coding application that threatened clients like Cursor. Claude Code outperformed both Cursor and Microsoft's GitHub Copilot in terms of revenue. Quinn Slack, CEO of the coding startup Amp, stated that "if there is only one winner in the race for AI models, it can dominate and crush its clients."

Officially, Anthropic justifies these limitations to prevent foreign adversaries or rival labs from enhancing their technology with its models. However, some developers fear that even basic features for AI applications could be affected, suspecting that Anthropic wants to keep the best technology for its own products.

Anthropic's rapid growth, with revenues increasing fivefold in five months, surpasses that of OpenAI and the combined 32 largest AI startups. This momentum echoes past antitrust cases involving Microsoft and Google, which used their dominance to favor their own applications at the expense of their partners.