Personal Finance Arrives in Claude (Beta)
Anthropic launched Personal Finance in beta on Nov. 5 for Pro, Max and Ultra users in the U.S. Connect bank and credit card accounts through Plaid, then ask plain-language questions like where your money went this month. Claude can flag duplicate charges, unexpected fees and rising bills.
It is read-only: Claude cannot move or invest money. It can see account names, last four digits, balances, credit limits and up to two years of transactions. Anthropic says there are no ads or sponsored picks, and Personal Finance chats are not used for training. Per the coverage, it is not built for investment advice.
Claude Enterprise Adds Security Controls
Anthropic announced more visibility and control for security teams on Nov. 5. The list: a security center (beta), generally available inference hooks, centralized admin management (beta) and new access controls. It also adds EDR sensor support for Cowork in beta.
The EDR piece is the follow-on to yesterday's Cowork monitoring news, now bundled into a wider enterprise security pitch.
Claude Marketplace Adds 25+ Partner Agents
On Nov. 4, Anthropic said Claude Marketplace is adding more than 25 partner agents and products across security, productivity and creative work. Eligible enterprise customers can apply committed Anthropic spend toward them.
That makes Marketplace a procurement shortcut, not just a catalog.
Watermarks and Message Threads on the Platform
Per the release notes, text from Claude Mythos 5, Opus 4.7 and Sonnet 4.6 now carries Anthropic's text watermark (Nov. 3). On Sonnet 4.6, requests with a non-default temperature, top_p or top_k may not carry it.
A summary of the notes also lists a Messages API message threads public beta (Nov. 2). Check the docs for the details before you build on it.
Claude Code Holds at 2.1.318
No new Claude Code release landed Nov. 5-6. The latest is 2.1.318 from Nov. 4, which adds login expiry details to claude auth status and auto-backgrounds long subagent shell commands after 3 minutes.
Reminder: fast mode for Opus 4.8 is removed Nov. 30. Status checks showed all systems operational, with the last incident on Oct. 28.
Game-Cloning Rule Draws Developer Pushback
Anthropic added two sentences to Claude Code's prompt on Opus 5.5 to steer it away from recreating someone else's specific game. Anthropic's Addy Osmani said on Oct. 30 that original, genre and own-asset games should keep working.
Developers say they also hit refusals on reverse engineering and software analysis. Some switched to Opus 4.8. Those reports are anecdotes, and Anthropic's confirmation covers only the game rule.
FlashProve: Proving Which Model Answered
A secondary report says Anthropic open-sourced FlashProve, a research prototype that uses zkSNARK proofs to show a response came from a specific model without revealing weights.
This comes from an aggregator listing, so confirm it on Anthropic's own channels before leaning on it.
Trust Is the Product Now
Read-only access, no ads, no training on chats, EDR hooks and admin controls. Anthropic's pitch this week is the same in consumer and enterprise: you can let Claude in because you can see and limit what it does.
The catch is the game-rule fight. Guardrails that block real work cost goodwill fast. Watch how Anthropic tunes them.