Machine-learning educator Santiago Valdarrama (@svpino) canceled his Claude Max subscription on July 29th, saying in a post on X that he would stop paying Anthropic while it lobbies the US government to restrict competitors.
Valdarrama praised Anthropic's models and Claude Code even as he announced the split. The attached cancellation notice said his Max access would expire on August 19th. The post had recorded more than 111,000 views in the snapshot supplied with it, giving Anthropic's policy dispute a direct consumer consequence from a developer who regularly teaches others how to build with AI.
Valdarrama runs ML School, a production-focused AI and machine-learning program that says more than 2,000 students have completed its cohorts since March 2023. He describes himself on his personal site as a computer scientist, machine-learning engineer and writer whose content reaches developers, data scientists and engineers. His background includes a master's degree in machine learning from Georgia Tech and engineering work on projects for IBM, Dell, Cisco and NextEra Energy, according to Real Python's biography.
That history makes the cancellation more consequential than a generic complaint about a chatbot. Valdarrama is part of the audience Anthropic cultivated through Claude Code: experienced developers who influence which models, coding agents and workflows other technical users adopt. His post framed switching providers as an ethical purchasing decision despite his continued admiration for Anthropic's products.
The claim and the record
Federal disclosures support Valdarrama's assertion that Anthropic has expanded its political operation. Anthropic spent more than $3.5 million on federal lobbying during the first half of 2026, including nearly $2 million in the second quarter, Axios reported. That exceeded Anthropic's $3.1 million total for all of 2025.
The disclosures covered export controls, cybersecurity, AI safety standards, Defense Department procurement, health care and energy policy. Anthropic lobbyists met with Congress and officials at the White House, Commerce Department and Treasury Department. Those filings do not establish that Anthropic sought a blanket ban on open-weight competitors.
The sharper dispute concerns the policies Anthropic supports and their likely market effects. Anthropic was absent from a July 24th industry letter signed by Nvidia, Microsoft, Meta, Google, OpenAI, GitHub, Hugging Face, Mistral, Replit, Y Combinator and dozens of other companies. The signatories urged policymakers to avoid premature restrictions on downloadable models, arguing that open weights lower costs, limit vendor lock-in and strengthen competition.
Anthropic CEO Dario Amodei answered the criticism on July 27th in a company post. He wrote that Anthropic had never advocated banning open-weight models as a category and called models without dangerous capabilities a public good.
Amodei instead backed tighter controls on advanced chips shipped to China, government action against industrial-scale model distillation and mandatory safety testing for sufficiently capable models, whether their weights are open or closed. He also argued that powerful open-weight systems can create additional cyber and biological risks because their safeguards can be removed and the models cannot be withdrawn after release.
That position leaves Anthropic separated from much of the technology industry on the practical question facing policymakers. Anthropic remains the only major frontier lab that declined to sign the open-weight letter, according to Axios, even after Google and OpenAI added their names.
Policy becomes a product problem
The conflict maps directly onto the economics of AI. Anthropic sells access to closed models and retains control over their operation, pricing and availability. Open-weight models allow developers to run systems on their own infrastructure, modify them and move workloads without depending on one API provider. Restrictions or expensive testing requirements could therefore fall hardest on smaller model builders while reducing price pressure on closed providers.
Amodei says commercial protection is not Anthropic's objective. Valdarrama and other critics are judging the policies by their potential effect rather than Anthropic's stated intent.
A viral cancellation remains anecdotal, and Anthropic's business impact cannot be inferred from one account. The reputational problem is already visible. Anthropic built unusually strong developer loyalty around Claude Code, yet its Washington agenda is forcing some of those users to weigh model quality against support for open competition.
Valdarrama's exit shows how quickly that calculation can turn into lost revenue. Anthropic's models kept his praise. Anthropic lost the subscription.