Scarlett Johansson, Cate Blanchett among 800 artists calling AI training 'theft'

Approximately 800 creators — including Scarlett Johansson, Cate Blanchett, R.E.M. and Vince Gilligan — signed an open letter accusing major technology firms of 'stealing' copyrighted works to train AI models and demanding ethical licensing partnerships. The move comes amid roughly 60 ongoing US lawsuits (and similar cases in Europe) alleging unauthorized scraping of books, images, music and other copyrighted content; companies argue fair use while artists warn of harms to livelihoods. The dispute raises legal, regulatory and reputational risk for AI platform providers and could increase licensing costs or constrain training datasets if courts or regulators impose limits; OpenAI has already faced a high-profile voice-rights backlash and paused a disputed voice model.

Analysis

Market structure: Copyright litigation shifts economic rents toward large IP owners (networks/studios/major labels) and away from unlicensed model builders. Winners: rights-rich media (DIS, WMG) and scale infrastructure suppliers (NVDA, MSFT, GOOGL, AMZN) that can negotiate licences; losers: small pure-play generative-AI startups, text‑to‑speech vendors and margin‑sensitive streamers (SPOT). Expect consolidation: incumbents with balance-sheet heft gain pricing power to buy/licence datasets, squeezing smaller entrants over 6–24 months.

Risk assessment: Tail risk includes a definitive adverse court ruling or injunction within 6–18 months that forces retraining/remove datasets or levies per-unit royalties, causing 5–25% one‑off writedowns for exposed model providers and +100–500bp margin pressure for streaming/content platforms. Short term (days–weeks) we’ll see headline-driven 3–8% volatility spikes in tech/media; medium term (3–12 months) settlements/licensing deals will reprice business models; long term (1–3 years) recurring royalty regimes could convert fixed data costs into variable COGS.

Trade implications: Tactical trades should long IP owners and infrastructure while hedging headline risk on hyperscalers. Favor 6–12 month directional exposure to DIS and WMG and maintain convex, capped exposure to NVDA to capture secular AI infra demand. Use options to buy downside protection on large-cap tech (MSFT/GOOGL) and targeted puts on small AI names (eg C3.ai) where liquidity allows.

Contrarian angles: Consensus assumes protracted pain for big tech; we view licensing as a monetisation pathway—large tech can internalise fees and pass costs to enterprise customers, entrenching their moats. The market may be underpricing the upside for rights owners (DIS, WMG) and overpricing legal existential risk for NVDA; history (music/streaming settlements) suggests eventual commercial deals, not permanent bans.

Key Decisions for Investors

• Establish a 1.5% portfolio long in NVDA via a 12‑month call spread (buy 10% OTM, sell 30% OTM) to capture continued AI infrastructure demand; target +30% return, trim at +50% or cut at -30%.
• Buy a 2% long equity position in Warner Music Group (WMG) and a 2% long in Disney (DIS) outright, horizon 6–12 months to capture licensing tailwinds; sell half on +40% or on any definitive ruling that limits licensing royalties.
• Implement a pair trade: long WMG (2%) vs short Spotify (SPOT) (1.5%) equal notional, expecting margin compression for streamers if licensing costs rise; rebalance if SPOT outperforms WMG by >15% relative in 90 days.
• Buy downside protection: purchase 3‑month ATM puts equal to 1% of portfolio on MSFT and buy 6‑9 month 10% OTM puts on AI (C3.ai, ticker AI) sized 1% to hedge headline/legal shocks; unwind on expiry or on definitive settlement within 6 months.