Anthropic is implementing stricter usage policies that effectively ban third-party integrations like OpenClaw from Claude starting April 4th. Under the new restrictions, users will no longer be able to use their standard Claude subscription limits when accessing the AI through alternative harnesses, forcing them to pay additional fees for the same service. This move signals a broader industry trend of AI companies exerting tighter control over their platforms, potentially fragmenting the ecosystem of AI tools that developers and users have built around major language models.
The timing of Anthropic's restrictions comes amid broader organizational upheaval in the AI sector. OpenAI announced that Fidji Simo, its CEO of AGI deployment, is taking medical leave, representing another high-level departure from the company's leadership ranks. These internal shifts underscore the pressures facing AI companies as they navigate rapid growth, competitive markets, and evolving leadership priorities around deployment strategy.
Meanwhile, Utah became only the second U.S. jurisdiction to authorize an AI system to independently prescribe psychiatric medications without physician oversight, raising critical questions about transparency and accountability in AI-driven healthcare decisions. State officials argue the system could reduce costs and address care shortages, yet physicians warn that the opaque nature of AI decision-making poses risks to patient safety. Together, these developments illustrate the sector's fundamental tension: as AI capabilities expand and integrate deeper into critical systems, questions about control, transparency, and appropriate human oversight remain unresolved.