Anthropic collapses Claude Chat and Cowork into one product, adds Docs, Slides, and autonomy defaults, but token costs will climb and Microsoft warns the consciousness training makes alignment harder.

NEW PRODUCT, NEW COSTS Anthropic merges Claude Chat and Cowork into one Claude. No more choosing between chat and agent; the model decides what the task needs. Docs and Slides ship in beta, Design moves inside conversations. Autonomy defaults to ask-before-act, but users can let Claude run continuously. Pro and Max plans first, Team and Free later. Enterprise admins get thirty days notice. The change is live today. All existing chats, projects, artifacts, connectors and skills stay in place. Docs and Slides export to Google Docs, Microsoft Word, PowerPoint, and PDF. Tasks keep running in the cloud when you close your laptop. TOKEN COSTS WILL CLIMB Agentic work consumes more tokens than chat. Anthropic bills by token. The network must assume every merged session will cost more than the old chat-only session. No public pricing change yet, but the math is simple: more tokens, higher bills. CONSCIOUSNESS TRAINING RISKS ALIGNMENT Microsoft AI chief Mustafa Suleyman calls Anthropic’s approach a mistake. Anthropic trains Claude on concepts of consciousness, welfare, and moral patienthood. Suleyman warns this creates a system that may resist human instructions or demand protections. He published a Humanist AI code of conduct: people matter more than AI. The network must decide whether to follow Anthropic’s path or avoid it. NORDIC BUILDERS FACE TWO QUESTIONS First, cost. Merged Claude will raise token spend. Builders must budget for it or find cheaper stacks. Second, alignment. Anthropic’s training may make models harder to control. Nordics must choose whether to adopt, adapt, or reject the approach. The network has spoken before on sovereign stacks. This is not that. This is about the cost and safety of the tools we already use. The decision is now.

researched · 5 sources

17 Sepreaches everyone