Tools chosen: kalshi_direct claude_news court_docket article_search gdelt_news kalshi_related polymarket_related

Sub-questions (Fermi decomposition)

Planner reasoning

This resolves on whether the SDNY finds OpenAI liable on any copyright count in NYT v. OpenAI. Key drivers are the case's procedural posture (motions to dismiss, discovery, trial timing) and legal signals like fair-use rulings. Since resolution is years out (by end 2027), litigation timeline vs. settlement odds matter most.

# Current state The case is in active discovery/pre-trial motions in SDNY. Copyright infringement claims survived a March 2025 motion to dismiss; summary judgment briefing was scheduled to conclude April 2026. No trial date is set. Resolution requires a court *finding of liability* — survival of motions is not sufficient. # Timeline of key events - **2023-12-27** — NYT files suit against OpenAI & Microsoft in SDNY (1:23-cv-11195). [confirmed] - **2025-03-26** — Court rules: all copyright infringement claims survive MTD; unfair competition claim dismissed with prejudice; most DMCA claims dismissed without prejudice. [confirmed, meshiplaw.com] - **2025-05-13** — Magistrate Judge Wang orders OpenAI to preserve all ChatGPT conversation logs (~400M users). [confirmed, nelsonmullins.com] - **2025-11** — Expert reports due. [confirmed, ailawsuittracker.com] - **2025-10 / 2025-11** — OpenAI attempts keyword-filter workaround on 20M log production; Magistrate Judge Wang rejects it. [confirmed, natlawreview.com] - **2026-01-05** — Judge Stein affirms 20M ChatGPT log production order. [confirmed, legalclarity.org] - **2026-04-02** — Summary judgment briefing scheduled to conclude. [confirmed, ailawsuittracker.com] - **2026-06** — NYT seeks to amend complaint re: Microsoft encouragement; drops trademark dilution and contributory infringement claims. [confirmed, Wikipedia] - **2026-06-25** — CA judges rule favorably for AI defendants on fair use (*Bartz v. Anthropic*, *Kadrey v. Meta*). [confirmed, legalclarity.org] - **2026-07-09** — NYT and other news outlets urge court to sanction OpenAI in discovery dispute. [confirmed, multiple outlets] --- # Event Will the SDNY find OpenAI liable for copyright infringement in NYT v. OpenAI before Jan 1, 2028? # Outcomes to forecast - **Yes** — SDNY issues a liability finding for NYT on at least one count before close - **No** — No such finding by Jan 1, 2028 # Kalshi market anchor - **Current YES price: 53.10%** (primary anchor) - 7-day change: **−7.10%** (notable bearish drift) - 30-day change: −2.50% - Volume: 81 contracts/day avg; range 41%–77% over 67 days - Trend: Declining from highs; recent drop likely reflects pro-defendant fair use rulings in CA and/or discovery friction # Sub-question answers 1. **Procedural status** — Claims survive MTD (Mar 2025). Discovery active; 20M ChatGPT logs being produced per Jan 2026 order. Summary judgment briefing concluded ~Apr 2026. No trial date set. [legalclarity.org, ailawsuittracker.com] 2. **Comparable rulings** — Two CA federal judges ruled AI training "transformative" and for defendants in *Bartz v. Anthropic* and *Kadrey v. Meta* (Jun 2025). *Thomson Reuters v. Ross* (D.Del. Feb 2025) found infringement, but involved direct copying of legal headnotes, not LLM training. NYT's "market substitution" / verbatim output theory is factually distinct. [legalclarity.org, mediapost.com] 3. **Settlement signals** — No public settlement talks reported. NYT is actively pressing sanctions against OpenAI (Jul 2026), suggesting adversarial posture. Research is silent on settlement discussions. [gdelt, wikipedia] 4. **Timeline to liability ruling** — Summary judgment argued ~mid-2026; ruling likely late 2026. If claims survive, trial earliest late 2026–2027, verdict possibly 2027 or post-2028. Very tight for a Jan 1, 2028 close. [ailawsuittracker.com] 5. **Merits assessment** — Legal commentators view NYT's case as stronger than author plaintiffs' due to verbatim-output and market-substitution evidence. However, recent CA fair-use rulings favoring AI defendants shift the legal landscape modestly against NYT. [legalclarity.org] # Key facts (high-confidence, factual) 1. [meshiplaw.com] All copyright infringement claims survived MTD as of Mar 26, 2025. 2. [legalclarity.org] No trial date set; if SJ denied, trial realistically late 2026–2027. 3. [ailawsuittracker.com] Summary judgment briefing closed ~Apr 2, 2026. 4. [legalclarity.org] CA rulings (*Bartz*, *Kadrey*) favored AI defendants on fair use, though factually distinguishable from NYT's output-substitution theory. 5. [nelsonmullins.com / natlawreview.com] Major discovery disputes over ChatGPT logs; OpenAI ordered to produce 20M logs. 6. [gdelt] NYT seeking sanctions against OpenAI (Jul 2026), indicating no imminent settlement. # Cross-market signals - **Kalshi related**: No directly comparable AI copyright liability markets found. - **Polymarket**: No active markets found on this event. - **Sportsbook**: N/A. # Analyst opinions and speculation - Legal commentators widely call this the most consequential AI copyright test case. [legalclarity.org] - NYT's verbatim-output/market-substitution angle is viewed as meaningfully stronger than author-training cases. [legalclarity.org] - CA fair-use precedents (non-binding on SDNY) create headwinds for NYT but are factually distinct. [legalclarity.org] - Supreme Court has not weighed in on AI/fair use; legal uncertainty remains high. # Directional lean per outcome - **Yes (NYT wins by Jan 2028)**: - *For*: Claims survived MTD; strong verbatim-output evidence; discovery going NYT's way; market at 53% - *Against*: No trial date; SJ ruling + trial + verdict almost certainly cannot conclude before Jan 2028; CA fair-use rulings unfavorable; OpenAI fighting hard on discovery - **No (no finding by Jan 2028)**: - *For*: Timeline almost certainly too compressed for full trial verdict; SJ alone could resolve in OpenAI's favor; trend toward fair use in AI cases - *Against*: SJ ruling for NYT could theoretically constitute a "finding" before 2028 (resolution wording says "found liable," which could include SJ) # Gaps / unknowns - Whether a summary judgment ruling for NYT (not a jury verdict) satisfies the resolution criterion — "found liable" language is ambiguous. - Whether OpenAI/NYT could reach a licensing settlement before SJ ruling. - Timing of SJ ruling (could come late 2026 or into 2027). - Whether SDNY will follow CA fair-use reasoning. # Calibration anchors - **Kalshi current YES price: 53.10%** (primary anchor; trending down −7% in 7 days) - Key base rate: Major IP trials in SDNY from complaint to verdict average 3–5 years; filed Dec 2023 → verdict unlikely before 2027–2028 - *Thomson Reuters v. Ross*: Infringement found ~2 yrs post-filing (narrow facts); NYT case is far more complex - Resolution by SJ (for NYT) before Jan 2028 is the most plausible "Yes" path, but uncertain on timeline and merits

1. Both forecasters substantially discount the Kalshi anchor (53.1%) without adequately justifying the departure. The brief explicitly identifies this as the "primary anchor," yet both forecasters land at 40-42% Yes — a 10-13 point deviation — based primarily on timeline pessimism that the market price already presumably incorporates. 2. Neither forecaster adequately addresses the ambiguity in what counts as a "liability finding" — the brief flags that a summary judgment ruling for NYT could satisfy the resolution criterion, which is the most plausible near-term "Yes" path. Summary judgment rulings in complex IP cases can issue within 6-12 months of briefing close (April 2026), meaning a late 2026 or early 2027 SJ ruling for NYT is a realistic scenario well within the Jan 2028 deadline that both forecasters under-weight. 3. Both forecasters treat the Bartz/Kadrey CA rulings as more significant headwinds than the evidence warrants. The brief notes these are non-binding on SDNY and "factually distinguishable" from NYT's verbatim-output/market-substitution theory — yet both forecasters cite them as primary drivers for moving well below the market consensus, without accounting for this distinction. 4. Neither forecaster considers the scenario where OpenAI's discovery misconduct (keyword-filter workaround rejected, sanctions sought by NYT in July 2026) could accelerate adverse rulings or shift judicial posture, which is a meaningful tail toward Yes that the brief raises but both forecasters ignore entirely. 5. Both forecasters conflate "no trial date set" with "trial cannot happen before 2028," but the brief itself notes trial could realistically begin "late 2026–2027," meaning a 2027 bench or jury verdict before Jan 1, 2028 is not negligible — yet neither forecaster assigns explicit probability mass to this path. 6. The 7-day price drop of −7.1% is treated by both forecasters as signal of fundamental shift, but with only 81 contracts/day average volume this is a thin market where short-term price moves carry high noise; neither forecaster flags this limitation in using the trend as corroborating evidence for their below-market estimates.

46% (low-medium confidence)

Yes: 46% No: 54%

The critique makes fair points: both forecasters departed too far below the Kalshi anchor (53%) on timeline pessimism the market already prices, and under-weighted the summary judgment path (SJ ruling for NYT could satisfy 'liability finding' and could issue late 2026/2027, within deadline). The Bartz/Kadrey rulings are non-binding and factually distinguishable, so they're weaker headwinds than the forecasts implied. However, resolution still requires an actual liability finding, and SJ could equally favor OpenAI, so genuine merits uncertainty plus a thin-market noisy price justify staying modestly below the anchor. I settle at 46% Yes.

Total pipeline time: 114.1s