New York Times reporter sues Google, xAI and OpenAI over AI training on copyrighted books

John Carreyrou and five other authors filed a federal copyright suit in California alleging xAI, Anthropic, Google, OpenAI, Meta and Perplexity used their books without permission to train large language models, with the plaintiffs explicitly not seeking class-action status. The complaint cites Anthropic’s August $1.5 billion settlement (noting class members would receive only about 2% of the Copyright Act’s $150,000 statutory maximum per infringed work), underscoring mounting litigation and liability risk for AI firms that may drive settlements, regulatory scrutiny and potential valuation impacts.

Analysis

Market structure: Copyright suits raise near-term bargaining power for book publishers and authors and create an incremental licensing market that could shift costs from near-zero training ingestion to recurring royalties; precedent (Anthropic $1.5bn) sets an implicit settlement band of $1bn–$5bn for deep-pocket AI firms, which is immaterial to Alphabet (>$100B cash) but could compress margins for smaller AI pure-plays by 1–3% of revenue over 12–36 months. Competitive dynamics favor vertically integrated incumbents that can absorb legal costs and negotiate licenses; smaller model vendors face higher unit costs and potential loss of funding, accelerating consolidation. Cross-asset: expect near-term equity vol upticks in big tech (IV +20–40% on headline days), modest widening of tech credit spreads (<20–30bps), limited FX/commodity impact, and safe-haven flows into treasuries on major adverse rulings.

Risk assessment: Tail risks include an injunction or statutory damage judgment aggregating into high-double-digit billions (theoretical maxes exist—$150k/work → catastrophic but low probability), or rules forcing provenance audits that add recurring capex. Immediate (days) risk = headline-driven share moves of 2–6%, short-term (weeks–months) = negotiating settlements or judge precedents, long-term (quarters–years) = durable licensing costs and product redesign. Hidden dependencies: class opt-outs, publisher consortium formation, and legislative responses could rapidly change liability exposure. Catalysts: court rulings in first-filed cases, large settlement announcements, or federal legislation within 3–12 months.

Trade implications: Tactical: favor long positions in cash-rich, diversified tech (Alphabet) on >3% selloffs while hedging with short-dated puts; selectively short unprofitable AI pure-plays and ad-dependent social platforms (Meta) that are reputationally exposed. Use 3-month put spreads to cap cost if IV spikes >25%, and pursue pair trades (long GOOG, short META) over 3–9 months to express relative resilience. Sector rotation: shift 20–40% of AI/midcap exposure into large-cap cloud/infra and content owners likely to monetize licensing (publishers, legacy media).

Contrarian angle: Market may overestimate perpetual liability—legal outcomes historically channelled into licensing (music, archives) not extinction; the Anthropic settlement suggests capped, negotiable outcomes rather than catastrophic damages. If courts limit statutory damages or favor fair-use defenses, large-cap AI equities could rebound 10–25% within 6–12 months, creating mean-reversion trades. Unintended consequence: stronger IP enforcement could create predictable royalty revenues for publishers and new B2B licensing platforms, a potential long idea ignored by consensus.

Key Decisions for Investors

• Establish a 1.5% long position in GOOGL (Alphabet) on any intraday weakness >3% and hold 6–12 months; target +12% upside while funding with 0.5% portfolio cash to buy 3-month ATM puts as hedge if IV >20%.
• Initiate a relative-value pair: long GOOGL 1.5% and short META 1.0% for a 3–9 month horizon to capture differential resilience to litigation and ad/reputation risk.
• Buy 3-month put spreads (buy 2.5% OTM, sell 7.5% OTM) sized 0.5% portfolio on META and on GOOG if headline IV rises >25%; objective: limit downside to headline-driven 5–15% moves while capping premium outlay.
• Reduce exposure to small/mid-cap pure-play LLM/Ai providers by 30–50% immediately and reallocate that capital into large-cap cloud/infra names and selected content owners; rationale: funding and licensing pressure will raise survival thresholds over 12–24 months.
• Establish a 0.5–1.0% long in NYT on any pullback >5% over the next 3 months to play potential publisher licensing upside; set a 12-month target of +15–20% contingent on licensing headlines or consortium formation.