Amodei urges US veto power over risky AI models as Anthropic faces Pentagon standoff
Anthropic CEO Dario Amodei says the US government should have authority to block developers from releasing risky AI models. He argues for “responsible scaling policies,” where the higher the risk of AI models, the stronger the required safety and testing before deployment.
Amodei has pushed this agenda since leaving OpenAI in 2021. In June 2025, he published an op-ed calling for federal transparency laws and risk mitigation requirements for AI companies. The framework gained momentum through state-level action, including California’s SB 53 (late 2025), which mandates safety testing and cybersecurity measures for frontier AI models, alongside New York’s AI regulations.
A separate conflict has intensified his message. In February 2026, Anthropic refused Pentagon demands to modify its Claude models for military use, including lifting certain safety safeguards. The refusal led to a temporary federal ban on Anthropic products, a supply-chain risk designation, and legal disputes. A partial resolution followed in March 2026, after a federal court restored some access to Anthropic’s models.
Notably, the article highlights that Amodei’s regulation path has no direct intersection with crypto or digital assets, despite broader industry interest in blockchain-based AI and tokenized compute markets.
Neutral
This is primarily an AI safety and US regulatory/defense procurement story, not a crypto-specific development. The article explicitly notes no intersection between Anthropic’s regulatory push and crypto or digital assets, which limits direct spillover into token flows.
For traders, the market impact is likely neutral: the biggest near-term effect is on sentiment around AI infrastructure and policy risk—rather than on major crypto assets. However, heightened government involvement in “frontier AI models” (including possible veto/ban powers) can marginally increase overall tech-sector regulatory uncertainty, which has occasionally coincided with risk-off moves in broader risk assets. Still, without explicit crypto links, any reaction in BTC/ETH is more likely to be second-order (macro/volatility) than fundamentals.
In the short term, expect limited direct trading catalysts for crypto. In the long term, if AI governance tightens globally, it may indirectly shape demand for compute and data services—an indirect tailwind or headwind to crypto narratives tied to decentralized compute, but the article provides no concrete project/coin exposure.