Overview
- A funding pact announced Friday between Alphabet and Anthropic commits up to $40 billion, with $10 billion wired now and the rest tied to performance targets at a roughly $350 billion valuation.
- The capital push follows new money from Amazon this week, with reports of $5 billion now and $20 billion later, as Anthropic lines up massive cloud and chip capacity to run its models.
- Anthropic keeps its experimental model, Claude Mythos, in closed trials with about 40 vetted partners, mostly in the United States, with the United Kingdom the lone foreign government given access.
- The company confirmed unauthorized users reached Mythos through an outside provider on the day of its announcement, and Bloomberg reporting cited less than 1% of the discovered software flaws as patched.
- Anthropic says Mythos found thousands of high‑severity bugs across major operating systems and web browsers, and multiple outlets report it also generated working exploit code without human help, which has drawn urgent reviews from central banks, U.S. financial authorities and European regulators.