An Anthropic researcher says he and Claude proved the Riemann hypothesis.
• Open since 1859
• Out as Lean code only
• No paper a human can read yet
• Not peer reviewed
The checker passes. He says he hasn't mastered all the math in it. Is that enough for you?
Computer use is now in beta on Claude Team and Enterprise plans (Nov 20):
• Claude sees your screen, clicks and types in apps you allow
• Off by default, an admin turns it on
• Docs: "no sandbox" between Claude and your apps
Would your IT allow it?
Claude Code 2.1.328 came out Nov 19. What I noticed:
• Auto mode now works with Opus 4.6 and Sonnet 4.6 on Bedrock, Google Cloud, Foundry
• mcp_tool hooks can block an action if the hook call fails
• zsh fix: "echo ====" no longer breaks Bash commands
Link in reply.
Anthropic starts trading today on Nasdaq as ANT.
• $83 per share
• 980M shares, about $81B at that price
• Deal expected to close Nov 20
I care less about the stock, more about what quarterly earnings do to Claude limits and pricing.
Better limits or tighter ones?
Claude Platform added linked billing on Nov 17 (beta).
• Your users pay for Claude usage from their own Claude org
• No API keys to collect, store or rotate
• They set a monthly spend cap per app
Warp, Kilo Code, Factory and Linear are on the launch list.
Anthropic is cutting its vulnerability disclosure window:
• 90 → 30 days for high and critical bugs
• 45 days for the rest
• Starts Jan 1, 2027
Their reason: a bug one model finds, another will likely find too.
Maintainers, is 30 days enough to ship a fix?
On Friday OpenAI launched ChatGPT Wallet. Its agents can now pay for things.
• US Pro subscribers only, for now
• Visa, Mastercard, Amex
• Agent gets a tokenized card, not your real number
• You approve every purchase
Would you let an agent check out with your card?
Anthropic published a guide on always-on agents (Nov 13). It names 5 building blocks:
• Event inbox
• Timers
• Memory
• Permissions
• Outbox
The agent sleeps most of the time and wakes up for runs.
Which of the 5 is the weakest part of your setup?
Anthropic IPO, per Reuters and Bloomberg:
• About 6x oversubscribed
• $70–$78 a share, 825M shares
• Up to $64.4B raised, up to $1.4T valuation
• Nasdaq listing as "ANT" expected Nov 19
Will a public Anthropic change anything for Claude users?
Claude Code 2.1.323 is out:
• Code Mode (experimental): Claude writes JavaScript to call its tools
• plugin.json can pin a Claude Code version range
• /security-review asks before running
• auto mode no longer starts looser than set
Which one are you trying first?