Grammarly Is Facing a Class Action Lawsuit Over Its AI ‘Expert Review’ Feature

A class action was filed in the Southern District of New York alleging Superhuman/Grammarly misappropriated names and identities, seeking damages in excess of $5 million on behalf of hundreds of writers (named plaintiff Julia Angwin). Superhuman has disabled the contested "Expert Review" AI feature and says it will rework the product; CEO calls the claims "without merit" and plans to defend. Risk is primarily reputational and legal for the company and could spur regulatory scrutiny or further litigation rather than immediate market-moving financial impact.

Analysis

This episode is a grafting point: right-of-publicity litigation against AI “persona” features creates a liability channel that is discrete from traditional data-privacy or copyright suits. Expect plaintiffs’ lawyers to treat names/likeness claims as low-friction, high-ceiling suits that can be pursued in multiple venues simultaneously; a single favorable ruling or settlement will rapidly recalibrate commercial economics for any consumer-facing product that monetizes modeled expert voices. Legal exposure is not binary — settlements, injunctive relief, and consent-file requirements all impose fixed and variable costs (upfront licensing, ongoing royalties, engineering to implement provenance flags) that scale worse for small players than for incumbents.

Practically, this favors vertically integrated platforms with deep balance sheets and enterprise-grade compliance tooling: they can absorb licensing pools and build provenance as a differentiator, while standalone AI-native consumer apps that positioned “expert agents” as a lightweight feature face either margin compression or the need to pivot to subscription or affiliate licensing. Downstream: content owners (publishers, brand portfolios, academic presses) now have leverage to extract licensing fees or to bundle “voice-as-a-service” deals — a new revenue line that could offset churn if executed quickly and contractually.

Time horizon: expect three waves. Short (weeks–months) — product shutdowns, PR costs, and defensive code patches; medium (6–18 months) — lawsuits, selective settlements, and nascent licensing agreements; long (18–36 months) — sector rationalization where compliant players win pricing power. Catalysts that would reverse the punitive trend are quick, low-cost standardized licensing frameworks or a clear appellate ruling narrowing publicity claims to narrowly defined commercial endorsements, but neither is imminent.

Key Decisions for Investors

• Long NYT (NYT) — 6–18 month horizon. Rationale: legacy publishers gain bargaining power to license “voice” and provenance services; target +20–30% if licensing pilot deals announced, downside -15% in ad-revenue shock. Position sizing: 2–4% of risk budget.
• Buy Alphabet 12-month ATM call spread (GOOG) — buy calls / sell higher strike to fund. Rationale: large cloud/AI incumbents will capture compliance-driven demand for hosted, provenance-backed agent services; expect 15–25% upside vs 10–12% downside if macro weakens. Use 1:2 risk/reward spread and cap size to 1–3% of portfolio.
• Long CrowdStrike (CRWD) or Zscaler (ZS) — 3–9 month calls. Rationale: surge in enterprise demand for data-provenance, model-audit, and governance tooling; payoff if corporate customers accelerate security spend. Risk: single-digit revenue miss could send shares lower; keep exposure to 1–2% of capital.
• Tactical tail hedge: buy short-dated VXX calls or small position in VXX (1–3 month) to protect against a litigation/tech-sector repricing shock. Rationale: sudden regulatory/legal shocks historically spike realized vol in AI/tech complex; cost small insurance premium relative to potential drawdowns.