Our Take
Anthropic is playing catch-up in legal AI with basic automation tools while established players like Harvey command billion-dollar valuations.
Why it matters
Legal AI is becoming a major enterprise battleground, but widespread AI-generated court failures show the sector is moving faster than quality controls can handle.
Do this week
Legal teams: audit your current AI usage policies before deploying new tools so you can avoid the sanctions hitting dozens of attorneys for AI-generated errors.
Claude gets legal plugins and MCP connectors
Anthropic launched new legal-focused features for Claude on Tuesday, expanding its Claude for Legal offering with plugins covering commercial, privacy, corporate, employment, and AI governance law. The tools automate document search, case review, deposition prep, and drafting tasks.
The company also released Model Context Protocol (MCP) connectors that integrate Claude with existing legal software including DocuSign for document management, Box for file search, and Thomson Reuters' Westlaw for legal research. These connectors are available to all paying Claude customers (per company announcement).
The move comes as legal AI competition intensifies. Harvey, which uses agentic AI for legal workflows, raised $200 million at an $11 billion valuation in March (per company reporting). Rival startup Legora raised $600 million in a Series D last month and launched a high-profile ad campaign featuring Jude Law.
AI legal failures are mounting in real courts
While AI companies court law firms with automation promises, practical deployment has created serious problems. Dozens of lawyers have been sanctioned for submitting AI-generated documents containing fabricated case citations and fake legal precedents. California issued its first fine against an attorney who used ChatGPT to draft an appeal filled with fictitious quotes last year (per state records).
Federal judges have also been caught using AI to draft rulings, drawing Congressional scrutiny. Meanwhile, AI-generated lawsuits are overwhelming court systems with poorly argued cases that legal experts describe as "legal slop."
An Anthropic spokesperson said legal sector pressure to adopt AI is separating early movers from laggards: "The firms and in-house teams that move are pulling ahead fast."
Focus on error prevention over speed gains
Legal practitioners should prioritize accuracy controls before deploying new AI tools. The pattern of high-profile sanctions shows that courts are becoming less tolerant of AI-generated errors, even when attorneys claim ignorance of the technology's limitations.
Document review and research tasks where Claude's new plugins focus carry lower risk than brief writing or case strategy. Start with these bounded applications while developing verification workflows for any AI-generated content that reaches clients or courts.
The MCP connectors for existing legal software like Westlaw represent safer deployment paths than standalone AI document generation, since they maintain connection to verified legal databases.