Claude: weekly briefing
8 Nov – 15 Nov 2026
Top stories
- Anthropic releases Claude Fable 5.2 and Mythos 5.2 models (Anthropic, 9 Nov) — Anthropic says the new Fable and Mythos models lead its earlier models on most capabilities. It also cut API prices to $8 per million input tokens and $40 per million output tokens.
- Cursor confirms breach let attackers push malicious Claude Code extension (BleepingComputer, 14 Nov) — Cursor said attackers used a key taken from Fireworks AI to distribute tampered Claude Code extensions through its marketplace on 29 October. It estimates up to 5,000 users may have received the malicious builds.
- Claude Code 'Don't Ask Again' feature exposes developer secrets publicly (iTnews, 8 Nov) — Apiiro found Claude Code's approval feature saved plain-text passwords and API keys in settings files across over 1,000 public GitHub repositories. Anthropic's Boris Cherny said the company is investigating, but proposed fixes may not fully solve it.
- Anthropic seeks up to $64.4bn in IPO at about $1.4tn valuation (Al Jazeera, 10 Nov) — According to its SEC filing, Anthropic plans to offer 825 million shares priced between $70 and $78 and launched its investor roadshow. Al Jazeera reported the valuation is below earlier estimates of $1.5tn to $2tn.
- Goldman Sachs limits Claude Code to its engineers (LavX News, 8 Nov) — Under October guidance, Goldman restricted Claude Code and GitHub Copilot to about 12,000 engineers, cutting off most traders and bankers. People familiar with the matter said a review of productivity and risk concerns informed the change.
