Scan: Will OpenAI pay a tort claim with more than $1 million in damages before 2028?

date	predicted	market price	confidence
2026-08-29	32%	26%	53%
2026-08-22	55%	60%	47%
2026-05-10	62%	70%	40%

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: kalshi_direct kalshi_related polymarket_related court_docket court_docket article_search claude_news gdelt_news code_execution

Sub-questions (Fermi decomposition)

What tort lawsuits (wrongful death, defamation, product liability, negligence) are currently pending against OpenAI, and what is their procedural stage and expected timeline to trial or settlement before 2028?
Has OpenAI already paid or agreed to pay any tort settlement or judgment exceeding $1 million, and would a confidential settlement count as 'paying tort damages' under Kalshi's resolution criteria?
What is the base rate and typical time-to-resolution for large tech companies settling or losing wrongful-death/product-liability tort suits (e.g., Character.AI, Meta, Snap, Tesla Autopilot cases)?
What is OpenAI's stated litigation strategy — does it fight tort claims to judgment or settle, and how do Section 230 / First Amendment defenses affect its exposure for chatbot output?
Are there any scheduled trial dates or mediation deadlines in the Raine v. OpenAI or related ChatGPT suicide/psychosis cases that fall before January 2028?
How do Kalshi and Polymarket price this and related OpenAI legal-liability markets, and has the price moved on recent lawsuit news?

Planner reasoning

This is a Kalshi question about whether OpenAI pays >$1M in tort damages before 2028. Key uncertainties are the pipeline of pending tort suits (wrongful death cases like Raine v. OpenAI, defamation suits, product liability claims), how fast such cases resolve (trial vs. settlement) and whether settlements count as 'paying tort damages' under Kalshi's rules, plus OpenAI's litigation posture. I'll anchor on the Kalshi price, pull docket status on key cases, and search news for settlements or judgments.

2. Research Parallel tool calls

tool	status	time	items	summary
kalshi_direct	OK	0.2s	1	## This Market's Kalshi Data Before 2028 (KXOAIDAMAGE-28) - Current price (probability): 20.00% - 7-day price change: -35.00% - 30-day price change: -51.00% - Average daily volume: 46 contracts - Price range: 20.00% - 71.00% - Data points: 27 days
kalshi_related	OK	6.5s	3	3 related markets / summaries. series KXOAIDAMAGE: 0 markets (skipped 1 no-signal) \| keyword 'OpenAI lawsuit': ok \| keyword 'OpenAI tort': ok \| keyword 'OpenAI settlement': ok
polymarket_related	OK	6.4s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'OpenAI lawsuit': 0 markets \| keyword 'OpenAI settlement': 0 markets \| keyword 'ChatGPT lawsuit': 0 markets \| keyword 'OpenAI damages': 0 markets
court_docket	FAILED	0.0s	0	tool unavailable (missing env var)
court_docket	FAILED	0.0s	0	tool unavailable (missing env var)
article_search	OK	6.7s	22	Fetched 22 unique articles across 4 queries (lookback=180d). 'OpenAI wrongful death lawsuit ChatGPT settlement': 10 articles \| 'OpenAI tort lawsuit trial date': 10 articles \| 'OpenAI pays damages lawsuit judgment': 10 articles \| 'ChatGPT suicide lawsuit OpenAI Raine': 10 articles
claude_news	OK	27.0s	15	Key findings: - 20+ tort suits pending against OpenAI, mostly wrongful-death/product-liability, filed 2025-2026 in California state court, now coordinated as a mass proceeding. More than 20 lawsuits are now pending against OpenAI over alleged ChatGPT-related harms, including wrongful-death
gdelt_news	OK	151.3s	20	GDELT: 20 articles across 3 queries (lookback=120d). 'OpenAI lawsuit settlement': 10 hits \| 'OpenAI sued ChatGPT death': error GDELT rate-limited after retries (429) \| 'OpenAI defamation lawsuit': 10 hits
code_execution	OK	51.3s	0	## Model Results: P(OpenAI pays >$1M tort damages before Jan 1, 2028) Setup: ~2.2 years remaining; per-suit annual hazard h = P(qualifying settlement/judgment in a given year \| still active); cumulative per-suit prob = 1–e^(–h·T); correlation modeled via Gaussian copula (shared factor ρ = MDL/b

3. Evidence Brief Sonnet 5 · 6998 chars

# Current state Kalshi YES price sits at 20% (down sharply from 71% high, -51% in 30 days), reflecting market repricing after realizing no trial/settlement is imminent. Resolution requires OpenAI to actually PAY >$1M in tort damages (judgment or settlement) before 2028 — merely being sued or facing pending litigation does not resolve YES. As of research date (~Aug 2026), OpenAI has paid $0 in tort damages and is contesting liability in all major cases; no trial dates are set. # Timeline of key events - 2025-04: Adam Raine, 16, dies by suicide after ChatGPT interactions (confirmed). - 2025-05-19: Walters v. OpenAI defamation case — summary judgment for OpenAI, no damages (confirmed). - 2025-08-26: Raine v. OpenAI filed in SF Superior Court (confirmed). - 2025-10: Amended Raine complaint adds intentional misconduct/punitive damages claims (confirmed). - 2025-11: OpenAI answers Raine complaint, denies liability, blames user misuse/TOS violation (confirmed). - 2026-02-10: Tumbler Ridge, BC school shooting; shooter allegedly used ChatGPT (confirmed). - 2026-02: SF Superior Court coordinates ~12 wrongful-death/product-liability suits into JCCP No. 5431 (confirmed). - 2026-04: Federal judge declines to halt a wrongful-death case, keeps parallel litigation on track (confirmed). - 2026-04-29: Tumbler Ridge families file 7 suits against OpenAI/Altman (confirmed). - 2026-05 to 2026-07: Multiple new suits filed (Texas overdose death, FSU shooting, New Brunswick suicide, Florida medical-advice case) — now ~19-20+ pending tort suits (reported). - 2026-01 (per Character.AI precedent): Google/Character.AI reach mediated settlement in Garcia case ~14-15 months after Oct 2024 filing, terms undisclosed (confirmed, analogous precedent). - 2026-06: Florida reportedly first state AG enforcement action against OpenAI (reported). - Mid-late 2026 (expected, not yet occurred): Section 230/product-liability motion-to-dismiss rulings across coordinated cases (rumored/expected per legal analysts). - No trial date set in Raine or any coordinated case as of latest research (confirmed). # Event Will OpenAI pay tort damages exceeding $1 million (via settlement or judgment) before January 1, 2028? # Outcomes to forecast - Yes - No # Kalshi market anchor Current YES price: **20%**. Sharp decline: -35% over 7 days, -51% over 30 days, off a high of 71%. Average daily volume only 46 contracts (thin market). This repricing suggests the market initially overestimated near-term settlement odds and has corrected as no trial dates or settlements materialized. # Sub-question answers 1. **Pending tort suits/timeline**: ~19-20+ wrongful-death/product-liability suits pending, mostly consolidated into JCCP No. 5431 (Feb 2026, SF Superior Court). Lead case Raine v. OpenAI (filed Aug 2025) has no trial date as of mid-2026; motions to dismiss on Section 230 expected mid-late 2026 but trials unlikely before 2028 [claude_news]. 2. **Prior payments/settlements**: No confirmed OpenAI tort payment >$1M to date. One defamation case (Walters) resolved via summary judgment for OpenAI — zero damages [claude_news]. Confidential settlement terms are common in comparable cases (Character.AI) and would likely count under Kalshi's criteria if damages exceed $1M, though amount may not be disclosed, creating resolution ambiguity. 3. **Base rate/precedent**: Character.AI/Google wrongful-death suit (filed Oct 2024) reached mediated settlement ~14-15 months later (Jan 2026), covering multiple states, amount undisclosed but "likely well over $1M" given severity [claude_news]. This is the closest analogue for AI-chatbot wrongful death liability speed. 4. **Litigation strategy**: OpenAI is fighting, not settling — denying liability in Raine, blaming user misuse/TOS violation, drawing public backlash [claude_news]. Plaintiffs use product-liability framing (Lemmon v. Snap precedent) to bypass Section 230 defenses. 5. **Scheduled trial dates**: None found before 2028 in Raine or coordinated JCCP cases as of research date [claude_news]. 6. **Cross-market pricing**: No Polymarket equivalent found (0 matches). Related Kalshi markets show no direct arbitrage signal; KXOAIDAMAGE-28 has moved independently, dropping from 71% to 20%. # Key facts (high-confidence, factual) 1. [Kalshi] Current YES = 20%, down from 71% peak; heavy recent selloff. 2. [claude_news/CBS/CNN/NPR] 19-20+ tort suits pending against OpenAI, coordinated in JCCP No. 5431 (Feb 2026). 3. [claude_news] Raine v. OpenAI filed Aug 2025; OpenAI denies liability; no trial date set. 4. [claude_news] Character.AI/Google settled analogous wrongful-death claims ~14-15 months post-filing (Jan 2026). 5. [claude_news] Walters v. OpenAI (defamation) resolved via summary judgment for OpenAI, zero damages, May 2025. # Cross-market signals - Kalshi related: Apple v. OpenAI IP case market at 50% (different tort type, before 2029) — not directly comparable. - Polymarket: No matching markets found. - Sportsbook implied: N/A. # Analyst opinions and speculation - Code-execution hazard-rate model (Monte Carlo across N suits, per-suit hazard, correlation): blended estimate ~87% before friction adjustment; point estimate after discounting for delays/dismissals/confidential-settlement ambiguity: **65-75%**, plausible range 55-90%. - Model flags key uncertainty: correlation across suits (single bellwether/global settlement could resolve many suits at once, lowering probability vs. independent-suit assumption). # Directional lean per outcome - **Yes**: Large volume of suits (20+), Character.AI precedent shows AI wrongful-death settlements can occur within ~15 months, ~2.2 years remain, product-liability framing survives some early motions, settlement pressure mounts with more filings/AG scrutiny. - **No**: OpenAI's stated strategy is aggressive litigation, not settlement; no trial dates set; Section 230/product-defect motions could dismiss cases before payment; confidential settlements might be structured under $1M or terms never disclosed enabling resolution ambiguity; Kalshi price has fallen sharply signaling market skepticism on near-term timeline. # Gaps / unknowns - No court docket data (tool failed) — actual procedural stage/deadlines unverified beyond secondary reporting. - Unclear how Kalshi resolves confidential/undisclosed settlement amounts. - No visibility into settlement negotiations, if any, in Raine or other lead cases. - Uncertain whether Section 230 motions will succeed, which is pivotal to case survival. # Calibration anchors - Kalshi current YES price: 20% (anchor, but volatile/thin market). - Character.AI/Google precedent: wrongful-death AI suit settled within ~15 months of filing — supports plausibility of pre-2028 resolution given multi-year runway. - Model-based estimate: 65-75% central, 55-90% range — notably higher than current Kalshi price, suggesting potential mispricing or model overconfidence given real-world litigation friction and OpenAI's fight-not-settle posture.

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

42% conf

Yes	32%
No	68%

Kalshi YES sits at 20% after a steep, thin-volume slide from 71%, reflecting that OpenAI has paid $0 in tort damages, is denying liability in Raine and the ~20 coordinated JCCP 5431 suits, and no trial dates exist before 2028. The main YES pathway is a settlement rather than a judgment: the Character.AI/Google wrongful-death case settled ~15 months after filing, and with 20+ suits, mounting AG scrutiny, and ~17 months of runway, a bellwether or global settlement by late 2027 is plausible. However, any settlement would likely be confidential, and the resolution requires a verifiable payment exceeding $1M, creating real ambiguity that cuts against YES even if OpenAI does settle. I lean modestly above the market given the volume of suits and settlement precedent, but well below the hazard-rate model's 65-75%, which underweights OpenAI's fight-not-settle posture, pending Section 230 motions, and disclosure friction.

gpt-5.6-sol

61% conf

Yes	28%
No	72%

I anchor on Kalshi's 20% YES price but raise it modestly because roughly 20 pending tort suits create multiple paths to a settlement, and the analogous Character.AI case settled about 14-15 months after filing. However, OpenAI is contesting liability, won the Walters case without damages, and has no trial scheduled before the deadline, all of which favor No. The suits are correlated through coordinated litigation rather than independent chances, while possible dismissals and confidential settlement amounts further reduce the likelihood of a verifiable payment exceeding $1 million. The brief's 65-75% model estimate appears to underweight these procedural and resolution frictions.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasts land at ~28-32% YES, essentially splitting the difference between the Kalshi 20% anchor and the model's 65-75%, but neither explains why the model's structured hazard-rate approach (which explicitly accounts for correlation across suits) should be discounted so heavily rather than partially trusted — they gesture at "fight-not-settle posture" without quantifying how much probability mass that actually removes from a 65-75% base.

Neither forecaster grapples with the resolution-ambiguity issue in a rigorous way: if a confidential settlement occurs but the amount is never disclosed, how does Kalshi actually resolve the market (YES, NO, or dispute)? This is treated as a vague "friction" that only pushes toward No, but it's equally possible the market resolves YES based on circumstantial/reported evidence (e.g., reported settlement size), which could cut the other way.

Both underweight the scenario where Section 230/product-liability motions to dismiss fail broadly (as flagged as "expected mid-late 2026" in the brief) — a wave of survived motions could dramatically increase settlement pressure well before 2028, yet neither forecast discusses how a favorable-to-plaintiffs ruling wave would shift their probability upward from current levels.

The Character.AI precedent (settlement ~15 months post-filing) is cited by both as supportive of YES, but neither adjusts for the fact that Raine was filed Aug 2025, meaning the 15-month analog point already passed (~Nov 2026) without a settlement — this is evidence the analogous timeline is NOT tracking for OpenAI, which should arguably lower confidence in near-term settlement, not just be mentioned as supportive precedent.

Neither forecast seriously addresses the tail scenario where a single catastrophic case (e.g., Tumbler Ridge mass-shooting suits, which involve multiple deaths and potent facts) produces an outsized punitive judgment or forces an early settlement specifically due to reputational risk, independent of the broader JCCP timeline — this could resolve YES faster than the "20 correlated suits" framing suggests.

Both forecasts move only modestly off the 20% Kalshi anchor (to 28-32%) despite flagging the model estimate at 65-75%, but neither justifies why the appropriate weighting should favor the thin, volatile Kalshi price (46 avg daily volume, -51% in 30 days with no new resolving information cited) over the more structured model — the anchoring rationale is asserted rather than argued.

6. Aggregator Fable 5.1 · final calibrated forecast

30% (52% confidence)

Yes: 30% No: 70%

Both forecasters converge near 28-32%, and the critique doesn't overturn their core logic: YES requires a verifiable >$1M payment, OpenAI has paid $0, denies liability across ~20 coordinated JCCP suits, and no trial is scheduled before 2028. The critique's point that the Character.AI 15-month analog has already 'passed' is factually wrong (Raine was filed Aug 2025; 15 months is Nov 2026, after the ~Aug 2026 research date), so it doesn't justify moving lower; conversely, the hazard-rate model's 65-75% ignores correlation through coordinated litigation and, critically, the likelihood that any settlement is confidential with an undisclosed amount, which may block YES resolution even if OpenAI pays. Decomposing: roughly 45-55% chance OpenAI settles at least one wrongful-death case by end-2027 (bellwether pressure, sympathetic facts, deep pockets), times ~55-65% that a verifiable >$1M figure surfaces, yields ~30%. The sharp Kalshi selloff on thin volume warrants some but not full deference, since it may reflect unreported negative procedural news, so I settle slightly above the 20% anchor at 30% YES.

Pipeline Timing

Total pipeline time: 256.4s

Per-tool research timings shown in the Research section above.