Publishers seek to join lawsuit against Google over AI training

Hachette Book Group and Cengage asked a federal court in California to intervene in a class action against Google, alleging the company copied their books and textbooks without permission to train its Gemini large language model and seeking unspecified monetary damages on behalf of authors and publishers. The publishers cited ten specific works (including authors Scott Turow and N.K. Jemisin) and say their participation could increase potential damages and bolster evidentiary issues; U.S. District Judge Eumi Lee will decide whether they may join. The move adds to a wave of high-profile copyright lawsuits over AI training—highlighted by Anthropic's $1.5 billion settlement with authors—and raises additional legal and financial risk for Google and other AI firms.

Analysis

Market structure: Copyright plaintiffs (large trade/academic publishers like Pearson/RELX and author classes) gain bargaining leverage; successful intervention and precedent will create a commercial market for licensed training data and royalties, shifting cost of model builds upward by an estimated 5–15% of training budgets within 12–36 months. Losers are primarily unprotected, data-hungry AI model vendors and ad/engagement-driven platforms (Alphabet GOOGL, Meta META) facing higher operating/legal costs and transient demand shock as compliance/redaction ramps.

Risk assessment: Tail risks include a multi-billion-dollar settlement or injunctive relief that forces retraining or licensing (range $1.5bn–$20bn; ~0.1–1% of top-tech market caps) — low probability but high impact for concentrated tech equity holders. Immediate (days) impact is elevated IV and headline-driven drawdowns; short-term (weeks–months) is litigation/news volatility; long-term (years) is structural licensing regime that could compress gross margins on foundation-model products by mid-single digits.

Trade implications: Tactical trades: buy underpriced exposure to publishers/academic content owners (RELX LSE: REL, Pearson PSON) as beneficiaries of licensing, and hedge with short-tail protection on big-model vendors (GOOGL, META). Use options around legal-catalyst windows: 3–6 month put spreads on GOOGL/META sized to 1–2% portfolio risk if implied volatility >35%, and buy covered calls or selective longs in publishers for yield enhancement.

Contrarian angles: Consensus treats this as headline noise; the market underestimates publishers’ ability to monetize licenses — a realized licensing market could add +5–10% steady-state EBITDA to top publishers over 2–3 years. Conversely, settlement sizes could remain immaterial vs. tech cash flows, creating short-term buying opportunities in tech names after volatility spikes rather than permanent structural declines.

Key Decisions for Investors

• Establish a 2–3% long position in RELX (REL.L) and/or Pearson (PSON.L) across the next 30 trading days, target blended entry 5–10% below recent levels, hold 6–18 months to capture licensing upside; take profits if combined position gains >25% or if court rules against publishers.
• Allocate 1.5% portfolio capital to a hedged trade: buy 3-month GOOGL 2.5% OTM put spreads (sell deeper OTM) sized to risk no more than 0.5% portfolio; enter if GOOGL 30‑day IV >35% or after judge’s ruling to allow intervention (likely catalyst 30–90 days).
• Implement a pair trade: long 2% SCHL (Scholastic SCHL) or PSON vs short 1–1.5% GOOGL (equal dollar) over 6–12 months to capture relative benefit to content owners; unwind if GOOGL outperforms publishers by >15% or if legal motions dismiss publisher claims.
• Sell 1–2 month covered calls (20% OTM) on select large-cap AI infrastructure names (NVDA, AMD) to harvest premium during headline-driven IV spikes; avoid directional exposure >3% in companies whose primary product relies on unlicensed scraped data.
• Monitor docket: if Judge Eumi Lee permits publisher intervention within 30–60 days, increase publishers longs by +50% and widen GOOGL/META put protection; if intervention denied, reduce downside hedges by 50% and selectively buy tech dip within 2 weeks.