Daily brief: AI warning, US-China AI claims, Iran crypto news

A CoinDesk daily briefing rounded up three cyber, AI and policy developments. An Anthropic researcher who worked on pretraining at OpenAI and Anthropic resigned, warning both labs are "racing straight to self-improving super-intelligence" and betting on systems many builders fear could end human civilization this decade.

• U.S. security agencies accused six Chinese AI firms, including Moonshot (maker of Kimi), of systematically extracting capabilities from American models rather than building them independently.
• Iran eased currency controls to let exporters move earnings home in crypto, using overseas revenue to fund imports directly and bypass official FX channels under U.S. sanctions.