Anthropic AI Updates: May 17, 2026

1. Anthropic publishes legal-industry deployment guide with three-phase rollout framework

Anthropic. A follow-up to last week’s legal launch, this Claude blog post lays out how in-house legal departments and law firms should roll Claude out across Chat (research), Claude Cowork (matter-level work), Microsoft 365 add-ins (Word, Outlook, Excel, PowerPoint), and the Claude Platform (custom apps). The guide cites a 2026 FTI Consulting report showing 87% of general counsel report generative AI use within their teams, up from 44% the prior year and 20% in 2023, and packages a three-phase adoption roadmap, the 12 practice-area plugins (Commercial, Corporate, Employment, Privacy, Product, Regulatory, AI Governance, IP, Litigation, and more), use-case maps by department, and IT-leader guidance on hosting, data retention, privilege protection, SSO, and audit. Anthropic also points to its own legal team’s contract redlining and marketing-review workflows as an internal case study. Source

2. Claude Code v2.1.143 adds plugin dependency enforcement, projected context costs, and worktree bypass

Anthropic. Claude Code v2.1.143, tagged on 2026-05-15, hardens plugin management: claude plugin disable now refuses to disable a plugin while another enabled plugin depends on it (and prints a copy-pasteable disable-chain hint), while claude plugin enable force-enables transitive dependencies. The /plugin marketplace browse pane now shows projected context cost as per-turn and per-invocation token estimates before installation. A new worktree.bgIsolation: "none" setting lets background sessions edit the working copy directly without EnterWorktree for repos where worktrees are impractical, the PowerShell tool now passes -ExecutionPolicy Bypass (opt out with CLAUDE_CODE_POWERSHELL_RESPECT_EXECUTION_POLICY=1), and background sessions now preserve the model and effort level after waking from idle. Shift+Tab in attached agent sessions now cycles through auto mode, and a runaway stop-hook loop ends with a warning after 8 consecutive blocks (override via CLAUDE_CODE_STOP_HOOK_BLOCK_CAP). Source