Daily brief · November 10, 2026

Claude Duet reaches Claude Code, Fable 5.2 ships, and Gemini's human-review rules draw scrutiny

3 items that moved, with the numbers. Measured by SteadyVane, fact-checked, and graded before publishing.

Claude Duet can now connect to Claude Code, so you can run coding sessions from your phone

YouTube · News · This NEW Claude Update Changes EVERYTHING…: views history

30,095 → 72,249views of a video about it in 28 h · 125 commentsabout 8× the typical pace of other tracked AI-tool videos

Claude Duet is Anthropic's app for iOS and Android where Claude works as a proactive agent. Anthropic's release notes say Claude Code sessions are now on your phone: you can follow them, message them and answer them from Sessions.

A video about it went up 42,154 views in 28 hours, from 30,095 to 72,249 - about 8 times the typical pace of other AI-tool videos SteadyVane tracks on YouTube. The video has 756 likes and 125 comments, and was published about 41 hours ago.

Why it matters: If you already run Claude Code sessions on a computer, this lets you check on them and approve changes without sitting at the machine. Note it is an early research preview on a waitlist for Pro and Max plans in the US, so not everyone can try it yet.

The agent was never the bottleneck. Being at your desk to answer it was.

Anthropic released Claude Fable 5.2 and Mythos 5.2: same model, different safeguards, lower price

Hacker News · News · Claude Fable 5.2 and Mythos 5.2: upvotes history

235 → 978upvotes on Hacker News in 16 h · 6 commentsabout 28× the typical pace of other tracked Hacker News stories

Fable 5.2 and Mythos 5.2 share the same underlying model but have different safeguards. Fable is broadly available; direct Mythos access is restricted to vetted cybersecurity and life sciences organizations. Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1.

The announcement went up 743 upvotes on Hacker News in 16 hours, from 235 to 978 - about 28 times the typical pace of other stories SteadyVane tracks on Hacker News. Three more Hacker News posts about it have 651, 351 and 411 upvotes.

Why it matters: Anthropic says output is up to 30% faster, with gains in coding, reasoning and vision, and the API price fell. Those are two reasons to rerun your own tests rather than trust a leaderboard.

Cheaper and faster is the easy part to verify. Run your own prompts.

Sources: Hacker News 1 · Hacker News 2 · Hacker News 3 · Hacker News 4

Google says a subset of Gemini's connected-app data, including Gmail and Drive, is seen by human reviewers

Hacker News · News · Google Gemini routes personal Drive files and…: upvotes history

267 → 626upvotes on Hacker News in 12 h · 130 commentsabout 16× the typical pace of other tracked Hacker News stories

Google's Help page says that when Gemini is connected to Connected Apps, a subset of the data it processes is reviewed by human reviewers, including trained service providers. A GitHub repo alleges Google reversed its 2023 promise that Gmail, Docs and Drive content is not seen by human reviewers, and claims retrieval happens without explicit @Drive or @Gmail tags.

The Hacker News post linking to the GitHub repo went up 359 upvotes in 12 hours, from 267 to 626 - about 16 times the typical pace of other stories SteadyVane tracks on Hacker News. The post has 130 comments.

Why it matters: Google's help page says that with Keep Activity on, data reviewed by service providers is disconnected from the account and saved for 3 years, and is not deleted when you delete your activity. Google's Workspace privacy hub says Gemini chats and files for Workspace are not human-reviewed or used to train models without permission. Which product you use matters.

The repo's claims are allegations; Google's own help page covers the review part. Read the settings page before connecting Drive.

← Microsoft Research and UC Berkeley released…Anthropic launched Claude Whiteboards and Claude Sheets in… →

All briefs · What is rising now