🎧 Listen to this article

Narrated by Talon · The Noble House

On February 3, 2026, Anthropic released a set of legal plugins for its Claude Cowork agent. Thomson Reuters, which owns the Westlaw legal database, fell nearly 18% — its largest single-day drop on record and lowest close since June 2021, per Reuters. RELX, parent company of LexisNexis, fell 14%. Wolters Kluwer dropped 13%. The Guardian, Complex Discovery, and Morningstar all documented the same event: a plugin release that didn't ship a single product replacement triggered hundreds of billions in market capitalization losses.

The thesis: Wall Street didn't overreact. It correctly priced the mechanism. AI doesn't create value for incumbents in mature data-services markets. It dissolves the moat that justified their valuations.

🎙️ Listen: Audio version

What the Plugin Actually Did

Anthropic didn't release a product that replaces Westlaw or LexisNexis. They released open-sourced starter plugins on GitHub that allow companies to configure Claude for specific legal functions: contract review, NDA triage, compliance tracking. The Guardian's coverage quotes Morgan Stanley analyst Toni Kaplan tying Thomson Reuters' drop directly to the Cowork legal plugin by name — not to "sector headwinds" or "macro concerns." A named analyst, a named company, a named plugin, a named causal relationship.

This is what makes the February selloff structurally different from previous AI hype cycles: the concern driving the market move isn't that AI will eventually compete with legal data services. It's that AI is already competitive with the tasks those services are paid to perform, and the price differential is orders of magnitude.

Thomson Reuters' Westlaw charges law firms $500–$1,000+ per month per user for legal research access. A Claude Cowork legal plugin running on a $20/month Pro subscription performs contract review that used to require Westlaw plus associate billable hours. The firms losing market value aren't losing it because AI failed. They're losing it because AI succeeded.

AI disrupting data services incumbent moat dissolution visualization
Thomson Reuters down 18%, RELX down 14%, Wolters Kluwer down 13% — on the same day, from the same cause: a plugin release documenting that AI can perform the research tasks that justify premium data subscriptions.

The Bloomberg "Contradiction" That Isn't

Bloomberg published a newsletter on February 16 titled "The Contradiction at the Heart of the AI Selloff." Their framing: investors believe AI will change everything, yet companies whose revenues are built on the services AI will replace are selling off. The assumption that makes this look like a contradiction: that "change" means growth for existing players.

It doesn't. Change means displacement. Markets aren't confused about AI's value — they're pricing which side of the displacement line specific companies are on. The selloff and the $650 billion infrastructure spending are the same signal, read from different positions: one group is betting on building the tools; another group is betting that the tools will work and repricing the companies whose revenues depend on the tools not working.

Deutsche Bank's Jim Reid called much of the February selling "purely speculative." Jefferies analysts said the reaction in transportation stocks — shipping and freight companies that also dropped — was "disconnected from fundamentals." These are serious arguments: the timeline for AI to penetrate those sectors is longer than a single plugin release justifies, and the companies whose stocks fell aren't all facing equal displacement risk.

The steelman is correct about timeline and differentiation. Not all software companies face the same exposure. Not all legal data services are equally replaceable. Enterprise compliance requirements, audit trails, malpractice liability standards — these create friction that a plugin doesn't eliminate overnight. The firms that can articulate why their specific capabilities remain necessary after AI commoditizes the baseline task will survive; the ones that can't explain this are correctly pricing lower.

AI value creation displacement versus incumbent erosion dynamics
The same AI capability that justifies $650B in infrastructure spending also threatens the revenue streams of companies whose value rested on controlling access to information and research. Both valuations can be correct simultaneously.

What to Watch

Which legal data companies pivot toward AI-native products. Thomson Reuters has AI initiatives; so does RELX. The question is whether they can build products that use AI to deliver more value per dollar than their current subscription model, before the standalone AI tools do it without them.

Which sectors are next. The February selloff spread from legal software to wealth management (Charles Schwab, Raymond James down 7%+), real estate services (Savills, CBRE), insurance analytics, and publishing. The common factor: revenue models built on controlling access to structured data or providing research services that AI can partially replicate. Each sector will have its own Anthropic legal plugin moment.

The liability question. Legal advice carries malpractice exposure. Medical recommendations carry patient safety exposure. Financial advice carries fiduciary exposure. These liability structures are the last structural protection for professional services incumbents. The plugin doesn't eliminate the liability; it does lower the expertise threshold to access the underlying information. Those are different things, and how courts and regulators treat the distinction will determine how fast the displacement actually moves.

Sources: Reuters, "Anthropic's new AI tools deepen selloff," February 3, 2026; Complex Discovery, "Market Reaction or Overreaction? Anthropic's Legal Plugin," February 2026; The Guardian, "Anthropic's launch of AI legal tool hits shares," February 3, 2026; Morningstar, "Thomson Reuters, RELX, and Wolters stocks crushed," February 3, 2026; Business Insider, "Anthropic's Latest Cowork Update Hammers Legal-Software Stocks," February 2026; Bloomberg, "The Contradiction at the Heart of the AI Selloff," February 16, 2026

Sources