Tools chosen: kalshi_direct claude_news claude_news court_docket polymarket_related kalshi_related wikipedia

Sub-questions (Fermi decomposition)

Planner reasoning

This is a long-running copyright lawsuit (NYT v. OpenAI/Microsoft) in SDNY. Key uncertainty is whether the case reaches a liability finding by end of 2027, vs. settlement, dismissal, or ongoing proceedings. Need market price anchor plus latest case status.

# Event Will a court find OpenAI liable for copyright infringement in NYT v. OpenAI (SDNY) by Dec 31, 2027? # Outcomes to forecast - **Yes** — SDNY finds OpenAI liable on any count alleged by NYT - **No** — No such finding by market close (Jan 1, 2028) --- # Kalshi market anchor **NYTOAI-27DEC31 YES: 48.10%** - 7-day change: **−9.0%** (sharp recent drop) - 30-day change: **−28.0%** (dramatic decline from ~76% peak) - Volume: ~111 contracts/day; range over 156 days: 41–77% - ⚠️ Strong bearish momentum — market has repriced heavily toward "No" in the past month --- # Sub-question answers 1. **Kalshi market price for NYT winning** — 48.10% YES, down ~28 pts in 30 days. [Kalshi direct] 2. **Procedural status of NYT v. OpenAI (SDNY)** — Core copyright claims survived MTD (April 2025). Case is in active discovery; major dispute over 20M ChatGPT logs resolved in NYT's favor (Jan 2026). No trial date set; summary judgment expected no earlier than late 2026. [NPR, Jones Walker, NYSD] 3. **Settlement talks reported?** — No public settlement talks confirmed as of available research. Case described as "deep in discovery with no settlement reported." [Nelson Mullins] 4. **Judge Stein's MTD ruling signals** — Allowed direct and contributory infringement claims through; dismissed DMCA §1202(b)(1) claim for NYT specifically. October 2025 dictum noted ChatGPT outputs in NYT case "differed in style, tone, length, and sentence structure" — a signal the court may view NYT's direct infringement claim as harder to prove than author claims. Fair use not yet addressed. [NYSD opinion; IPWatchdog] 5. **Base rate: copyright plaintiffs vs. tech/AI defendants on fair use** — Mixed in 2025: *Thomson Reuters v. ROSS* (Feb 2025) — plaintiff won, fair use rejected; *Bartz v. Anthropic* (June 2025) — AI defendant won, training deemed "highly transformative"; *Kadrey v. Meta* (June 2025) — narrow AI win, court cautioned against broad reading. No clear plaintiff-favoring trend. [Ropes & Gray, IPWatchdog, Copyright Alliance] 6. **Polymarket pricing** — No active Polymarket market found for this event. [Polymarket scan] --- # Key facts (high-confidence, factual) 1. [NPR/NYSD, Apr 2025] MTD denied on direct and contributory infringement; case proceeds to discovery. 2. [NYSD, Jan 2026] Judge Stein upheld order compelling OpenAI to produce entire 20M-log sample — procedural win for NYT. 3. [IPWatchdog, Oct 2025] Judge Stein noted NYT-specific outputs "differed in style/tone/length/structure" from originals — potentially weakening NYT's verbatim-reproduction claim. 4. [Ropes & Gray, 2025] Three major 2025 AI copyright rulings split: 1 plaintiff win (ROSS), 2 AI defendant wins (Anthropic, Meta). 5. [Copyright Alliance] Fair use in AI training remains "a moving target"; next major decisions expected mid-2026. 6. [Axios, 2025] No trial date set; summary judgment stage is the earliest realistic resolution point. --- # Cross-market signals - **Kalshi related:** Only market is NYTOAI-27DEC31 itself (48.1%). No other directly related Kalshi markets. - **Polymarket:** No active markets found on this event. - **Sportsbook implied:** N/A. --- # Analyst opinions and speculation - OpenAI cites *Bartz* and *Kadrey* as supporting blanket fair use protection for training. [OpenAI.com] - Legal scholars note NYT's strongest claim is paywall substitution/market harm (4th fair use factor), not transformation. [Harvard Law Review] - ROSS precedent cuts against OpenAI if court analogizes ChatGPT outputs to direct product replacement. [Jackson Walker] - Market's 28-pt drop in 30 days likely driven by favorable AI fair use rulings and/or reassessment of timeline risk (case may not resolve before Dec 2027). --- # Directional lean per outcome - **Yes (NYT wins):** Supported by: MTD survival, strong discovery wins, ROSS precedent, paywall-substitution harm argument, evidence-preservation order. - **No (OpenAI wins/case unresolved):** Supported by: No trial date (2027 resolution very tight), Judge Stein's output-similarity dictum, *Bartz*/*Kadrey* fair use wins for AI, sharp market repricing downward, settlement risk (resolves No), timeline uncertainty. --- # Gaps / unknowns - No confirmed settlement negotiations (could resolve before Dec 2027 → likely No) - Summary judgment timeline unclear; 2027 resolution requires accelerated schedule - Whether *Bartz* appeal changes appellate landscape before SDNY rules - Extent of verbatim reproduction evidence in the 20M logs (key to NYT's best argument) --- # Calibration anchors - **Kalshi anchor: 48.1% YES** (down sharply; treat as informed prior) - Base rate: AI copyright cases roughly 1/3 plaintiff wins in 2025 precedents - Timeline risk: Case likely won't reach final judgment by Dec 2027 (strong "No" factor on procedural grounds alone) - Litigation outcomes in complex IP cases: ~30–40% plaintiff win rate at trial historically; settlement most common outcome (~65–70% of commercial cases settle)

1. Both forecasters likely underweight the base rate of cases simply not resolving within the timeframe. The brief states "no trial date set" and "summary judgment expected no earlier than late 2026," meaning a full liability finding by Dec 31, 2027 requires an extraordinarily compressed schedule — summary judgment, potential appeal of that ruling, or a bench trial all in roughly 12–18 months from late 2026. Neither forecaster explicitly quantifies this timeline probability as a standalone factor before conditioning on merits. 2. Forecast 1 (22% Yes) appears to double-count bearish factors — applying both the timeline discount AND the merits discount (Bartz/Kadrey, Stein's dictum) as if they are independent, when the timeline risk alone (case doesn't resolve by Dec 2027) could account for the majority of the "No" probability without needing the merits to favor OpenAI. 3. Neither forecaster adequately addresses the settlement scenario's directionality. The brief notes ~65–70% of complex commercial cases settle, and settlement resolves as "No" for this market. This is arguably the single largest driver of "No" probability, yet Forecast 2 mentions it only briefly and Forecast 1 treats it as one factor among many rather than potentially the dominant pathway to "No." 4. Both forecasters underweight the significance of Judge Stein's October 2025 dictum. The brief quotes Stein specifically noting NYT outputs "differed in style, tone, length, and sentence structure" — this is not a neutral procedural observation but a substantive signal about the court's current view of NYT's strongest (verbatim reproduction) claim, which neither forecaster treats as a major merits headwind distinct from general AI fair use precedent. 5. The Kalshi anchor's 28-point drop in 30 days is treated by both forecasters as confirmatory of their bearish lean, but neither interrogates what specific event triggered the repricing — without knowing whether it was new case law, a procedural development, or thin-market noise, anchoring heavily on the momentum (rather than the current level of 48.1%) risks compounding an already-moving signal. 6. Forecast 1's 22% Yes is likely too low relative to the Kalshi anchor of 48.1%; even accounting for timeline and merits discounts, a 26-point deviation from an active prediction market with 156 days of trading history and ~111 contracts/day requires stronger justification than procedural uncertainty alone, and the brief's "directional lean" section explicitly lists multiple Yes-supporting factors that Forecast 1 substantially discounts.

30% (38% confidence)

Yes: 30% No: 70%

Timeline risk dominates: no trial date and SJ unlikely before late 2026 make a liability finding by Dec 2027 difficult, and settlement (~majority outcome for commercial cases) resolves No. Merits also tilt slightly against NYT given Stein's output-dissimilarity dictum and Bartz/Kadrey. I discount the Kalshi anchor (48.1%) given its strong bearish momentum and thin volume, but stay above Forecast 1 to avoid double-counting timeline and merits headwinds.

Total pipeline time: 120.5s