Anthropic opened Claude Security scans to every enterprise customer Friday, and the scanning engine now running underneath is Mythos 5 — the company's most capable model, previously confined to a small vetted cohort. [1][2] Point the scanner at a GitHub repository and it returns findings tagged by weakness category, with confidence ratings, severity ratings, and a suggested patch that opens in Claude Code for human review. The model itself stays behind the scan; billing runs at standard token usage under existing plans. [1] A companion Defender Advantage Fund puts $35 million in credits behind open-source vulnerability patching. [1]

The same week added two more entries to the ledger. Anthropic hired a Google chip veteran for its custom-silicon push. [2] And a Broadcom-backed vehicle negotiating $70 billion to $80 billion in debt exists substantially to finance AI chip purchases — including Anthropic's. [3] SiliconANGLE's Friday tally argues the buildout has grown too large to unwind quietly, with OpenAI trailing Anthropic on growth and listing momentum. [4]

There is an American story here beneath the product notes. Enterprises do not adopt scanners; they adopt relief — a machine that reads the code nobody owns anymore. The company selling that relief is assembling its own silicon, its own financing, and a date with the public markets.

-- MAYA CALLOWAY, New York