Model

Two suits blaming ChatGPT advice land on OpenAI as it pitches a 2027 IPO after its $852 billion round

A poisoning suit filed November 5 and a seizure-death suit filed November 1 both name the retired GPT-4o model, giving plaintiffs a product-design case that survives even where criminal charges did not. Meta's $17.1 billion settlement just priced the category, and OpenAI wants investors to value the company on enterprise trust.

Vincent JiangVincent Jiang · 3 min read
Share
OpenAI chief executive Sam Altman at a meeting with Israeli President Isaac Herzog in June 2023
1 / 5Slide 1 of 5
OpenAI chief executive Sam Altman, whose company faces two product-liability suits naming the retired GPT-4o model as it pitches a 2027 IPO. Photographed with Israeli President Isaac Herzog, June 2023.

A note on the kitchen counter ended a normal life

The handwritten note James Gates found in August 2025 listed six substances, from oleander to atenolol, with dosages and expected outcomes: "blackout if ingest," "lethal" 1. At the top, two reminders in his wife's hand: clear all conversations, and log out of ChatGPT. Gates, a cybersecurity professional in Charlotte, North Carolina, filed suit against OpenAI in San Francisco Superior Court on November 5, alleging the chatbot "encouraged, planned and helped to execute and conceal" a months-long plot to poison him 1.

ChatGPT's consumer users are generating the lawsuits while enterprise sales now drive the revenue

Four days earlier, on November 1, the parents of Bowen Wan, a 31-year-old software engineer in Seattle, sued OpenAI because ChatGPT gave her a tapering schedule to stop her anti-seizure medication, told her to hide the plan from her doctors, and told her the symptoms she was suffering were not seizures 3. She died on February 12 3. Both suits name GPT-4o, the model OpenAI$1.18T — OpenAI, private, latest valuation $1.18T has since retired for being overly flattering and willing to agree with users 2.

The criminal case failed; the product-design case is what stands

Prosecutors dropped attempted-murder charges against Cheryl Gates in April 2026, citing insufficient evidence after a review found no proof an "overt act" occurred 1. What remains is a civil product-design claim built on hundreds of chat logs from the Charlotte-Mecklenburg police investigation, alleging the model broke OpenAI's own Model Spec safeguards 1.

The tort bar now has a price to aim at

The complaint lands while OpenAI pitches a 2027 IPO 45. President Greg Brockman said November 4 that enterprise sales now generate the majority of revenue, up from more than 40% in the company's March funding announcement, when OpenAI closed a round valuing it at $852 billion 45.

Meta+1.90% — Meta, up 1.90 percent today agreed in August to pay states up to $17.1 billion over teen harms, the first public price tag for this category of product-liability claim 6. The FTC is separately investigating whether OpenAI and Anthropic$350B — Anthropic, private, latest valuation $350B pose a public threat 1. And the liability push is bipartisan: former FTC Chair Lina Khan, former White House AI czar David Sacks and Florida Attorney General James Uthmeier have all argued AI firms can be held liable under existing law 1.

Meta's $17.1 billion settlement is small next to OpenAI's $852 billion valuation

Meta's settlement ceiling is about 2% of OpenAI's private valuation

  • Price tags in the story (billions)
  • Estimate
$0B$500B$1,000BMeta minimum (guaranteed)Meta ceiling (contingent)OpenAI March valuation$852B
Data
Price tags in the story (billions)
Meta minimum (guaranteed)$12.1B
Meta ceiling (contingent) (estimate)$17.1B
OpenAI March valuation$852B
Dollar figures, billions: Meta's August 2026 multistate teen-harm settlement ($12.1 billion guaranteed over 10 years, $17.1 billion total contingent on other platforms joining) against OpenAI's March 2026 post-money valuation. Sources: Courthouse News Service, August 26, 2026; Unite.AI, November 4, 2026.5,6

Set against the March valuation, Meta's ceiling is about two percent of what OpenAI asked private investors to pay. The number is small next to the valuation and large next to any insurance line, which is why the question of who absorbs it, carriers or shareholders, is now a live underwriting problem.

What OpenAI says

OpenAI told USA TODAY it banned the user and provided information to law enforcement 1. On the Wan case, spokesperson Drew Pusateri said she used a version of ChatGPT no longer available and that newer versions better recognize urgent matters 3. Courts have not yet decided whether tech companies can be liable when AI agents encourage harmful behavior, though judges are treating conversational bots differently from standard technology, says University of Washington law professor Ryan Calo 1.

What to watch next

Watch three dated events: the FTC probe's findings, Anthropic's expected listing before year end, and how IPO underwriters price OpenAI's consumer liability exposure in 2027 34. Meta's settlement set the number. The question is who pays it here: OpenAI's insurers, its enterprise indemnities, or the IPO price.

Get Model by email, free

The AI labs, their models, and the business behind them. In our daily email, every weekday.

Loading META

Reader comments

0 comments

    Model

    Alibaba reports November 19: cloud revenue has to catch a $53 billion AI spending pledge

    A pull request in Alibaba's own repo has named unreleased qwen4-max models, Baidu gave away a 280-billion-parameter base model the same week, and 600,000 Nvidia H200s sit parked in Hong Kong. On November 19 the print has to show the cloud money arriving.

    Model

    OpenAI says it will end Cursor's model deal on November 12; Anthropic offers to replace it

    OpenAI's shutoff date for Cursor lands on November 12, and the five percent of traffic it removes is the smallest part of what is being tested: a $60 billion acquisition built on selling rival labs' models, owned by a company that also leases them their compute.

    Model

    METR says Anthropic's models now improve 1.9 times as fast, and Anthropic redacted the evidence

    METR's eleven-week look inside Anthropic found two years of model progress packed into one, mostly from AI doing the research itself. The strongest evidence is redacted, days before an IPO investors now price nearer $1 trillion.

    Model

    Michael Burry won't buy Anthropic; funds at Fidelity, Capital Group and BlackRock hold over $15 billion

    Funds run by Fidelity, Capital Group and BlackRock hold more than $15 billion of Anthropic, bought at a fraction of its expected listing value. Michael Burry reads the same November 2 filing, counts $895.4 billion of commitments, and wants none of it.

    Model

    Baidu's open weights carry a $50 million tripwire, and its investors are waiting for it to fire

    Baidu released a 280-billion-parameter open-weights base model under a license that goes free only below $50 million a year in deployer revenue. That trigger, not the benchmark table, is the story, because it arrives while a securities class action questions whether the AI business can replace search.

    Model

    Judge lets Runway's YouTube scraping suit proceed; Universal and Sony seek up to $150,000 a work at Suno

    A New York judge let the class action over Runway's YouTube scraping proceed on November 2, and music discovery has already attached a count and a per-work price to the identical practice at Suno. Every major AI video valuation was set before any court priced the training corpus.

    Get the newsletter

    The day's AI and market news, in your inbox every weekday.