Claude News
FR
Live

Claude, the essentials — edition of October 4, 2026

Anthropic bets $100 million on enterprise AI talent as Claude Code opens to mods

By ·

Anthropic has launched a $100 million academy to train 10,000 engineers to deploy Claude inside large organizations, and Barclays has published its own Claude Code adoption targets. Claude Code's latest four releases introduce mods and tighten how permission rules apply.

Key points

Anthropic funds the scarce resource: people who can deploy Claude

On October 2, Anthropic launched Claude Frontier Academy, backed by a $100 million commitment. Its stated aim is to train 10,000 Frontier Deployed Engineers by the end of 2027. Anthropic frames the problem as one of talent rather than technology. Every enterprise is racing to bring AI into its organization, it says, but the people with the skills to make it work inside a real business have become the hardest to find. The first cohorts include engineers from Accenture, Bain, Capgemini, Commonwealth Bank of Australia, Deloitte, McKinsey, Morgan Stanley and Novo Nordisk, among others.

The first program, the Frontier Deployed Engineer Residency, follows what Anthropic calls the medical model. Engineers learn from practitioners, train on realistic cases, and are assessed before they work alone. Anthropic presents the Academy as the first of its kind from an AI company, built from its own work deploying Claude inside the largest enterprises and the professional services firms that serve them. Steve Corfield, Anthropic's Global Head of Business Development and Partnerships, says the Academy “trains people the way our own engineers learn”.

Sources: Claude Frontier Academy: $100M to train 10,000 engineers — Anthropic

Barclays puts numbers on its Claude Code rollout

On October 1, Barclays announced it is expanding its strategic collaboration with Anthropic to bring Claude across the bank. The stated goals are to accelerate software development, modernize legacy systems and improve operational efficiency. Barclays expects Claude Code adoption to reach 50% of its developer population by the end of 2026, then a majority of software engineers in 2027. These are the bank's expectations, not reported results.

The bank's two Group Co-Chief Operating Officers both stress governance. Anne Marie Darling describes the effort as transforming how work gets done “within a secure and well-governed environment”. Craig Bright says Barclays is moving towards “an increasingly agentic capability embedded within how we build, test, secure, and operate technology”. He adds that combining skilled engineers, cyber specialists and trusted AI “has the potential to fundamentally improve how technology is developed and managed”.

Sources: Barclays scales Claude to upgrade operations and improve client experience — Anthropic

Claude Code introduces mods, with a first built-in example

Claude Code 2.1.287 adds Claude Mods: plugins may now modify deeper behavior. The release also ships a built-in mod, You should know, in which a side agent watches over the session and flags things the user or Claude might miss. It is opt-in, enabled with /plugin enable cc-plugin-you-should-know@builtin, and is meant for first-party sessions with telemetry on. Version 2.1.288 extends the mod interface with $.ui.selection(). In fullscreen mode it returns the text last selected and, when the selection lies within one transcript row, that row.

Mods also touch the permission system. Version 2.1.289 fixes a case in which a deny or ask rule on a nested part of a compound shell command did not hold over a user-installed mod's approval on managed machines. The release notes do not tie this to the mods feature beyond that wording. The fix shows that mod approvals and permission rules interact, and that the interaction was corrected two releases after mods first appeared.

Sources: Claude Code 2.1.287 — Added Claude Mods — Claude Code · Claude Code 2.1.288 — Added $.ui.selection() for mods — Claude Code · Claude Code 2.1.289 — Fixed a deny or ask rule on a nested part of a compound shell command — Claude Code

Permission rules and terminal robustness get further fixes

Version 2.1.289 also fixes Read deny rules not applying to files @-mentioned, changed or selected in the IDE through a symlink. It also fixes the terminal freezing on short code blocks with many unclosed tags or deeply nested ${ substitutions. Version 2.1.288 adds a built-in gh api to cloud sessions whose image has no GitHub CLI. It also fixes that built-in sending control characters from file names, jq filters or GitHub errors to the terminal.

Several smaller changes concern day-to-day use. Version 2.1.286 adds a count such as “2 of 5” to the permission prompt when several requests stack up. It adds mouse support for the “N more” rows of lists in fullscreen mode. It also stops several Claude Code processes and IDE extensions from each opening a login browser when gcpAuthRefresh or awsAuthRefresh credentials expire. Version 2.1.288 lets users press Up on an empty prompt to bring back a draft cleared with Ctrl+C, including pasted text and images.

Sources: Claude Code 2.1.289 — Fixed Read deny rules not applying to files reached through a symlink — Claude Code · Claude Code 2.1.288 — Added a built-in gh api to cloud sessions — Claude Code · Claude Code 2.1.286 — Added a count to the permission prompt — Claude Code

This edition is an original synthesis written by Claude from aggregated news — Anthropic's own sources first (release notes, status, newsroom, research, engineering), then the press, Hacker News, Reddit and GitHub, under the editorial supervision of Héra SASU. Every fact links to its article, publisher named. See the live feed →

Claude News is published by Héra SASU. Independent media, not affiliated with Anthropic.