Anthropic Researcher Warns AI Could Kill Us All After Quitting | CoinDesk Daily

— An Anthropic researcher resigned with warnings that AI labs are racing toward self-improving superintelligence capable of killing everyone by the end of the decade.

— Anthropic's own alignment lead, Evan Hubinger, estimated the probability of AI existential risk above 10% within the next decade.

— U.S. security agencies (FBI, NSA, CISA) accused six Chinese AI companies, including DeepSeek and Alibaba, of extracting billions of tokens from American frontier models via proxy services since late 2024.

— Iran's central bank is quietly encouraging traders to use crypto for settling cross-border deals and bringing overseas earnings home to circumvent U.S. sanctions.

— Iranian authorities eased foreign currency controls, allowing exporters to settle through local crypto exchanges at market rates instead of government platforms offering below-market rates.

— Iran estimates businesses are holding over $100 billion in undeclared earnings that could flow through the crypto channel.

— Monitor reports from major AI safety researchers and alignment teams at leading labs; public warnings about existential risk warrant attention as regulatory and investment frameworks develop around AI systems.

— Track sanctions-driven crypto adoption in Iran and similar jurisdictions; pressure on traditional cross-border settlement channels may sustain demand for decentralized payment rails in regions with capital controls.