Unscrolled.
Live
entertainmentOscar-winning Chilean director Sebastián Lelio dies at 52businessOracle outranks Microsoft, Amazon on Wall Street valuation metrictechMan sues OpenAI after wife allegedly used ChatGPT to plan poisoningworldArtemis II launch created massive holes in Earth's atmosphereworldNithya Raman defeats incumbent Karen Bass in LA mayoral racesportsSaudi-Backed LIV Golf Shuts Down After Five YearsworldMissing Philadelphia girl found dead in New JerseybusinessWendy's Rejects $1.4B Buyout Bid, Adopts Poison Pill DefenseworldSpeaker Johnson Won't Seek Minority Leader Role After GOP LossesworldWest Point cadet pleads guilty to drugging girlfriend, later attends Ivy LeagueworldDemocrats gain secretary of state control in battleground statesworldAmericans vote in battle for Senate and House control— live updatesworldTennessee murderer captured over a week after escaping prisonbusinessSnap settles social media addiction lawsuit without trialworldChili’s customers suffer burns after sipping margaritas mistakenly made with cleaning agentworldEbola outbreak spreads across Africa; officials warn global threatworldWashington reports first US bird flu death of 2026worldFederal judge dismisses Trump administration's Alaska Senate lawsuitworldKing Charles to address slavery legacy in Antigua as PM opens to apologyworldIran's supreme leader warns U.S., Israel against resuming conflictworld281 arrested in Florida human trafficking and child predator stingworldUkrainian evacuees reunite with family after years trapped in Russia-occupied city - BBCbusinessNasdaq plunges 970 points as tech earnings disappoint investorsworldRose-colored Sahara meteorite confirmed as 4.1-billion-year-old Mars fragmentbusinessTech stocks plunge as AI spending concerns mountentertainmentOscar-winning Chilean director Sebastián Lelio dies at 52businessOracle outranks Microsoft, Amazon on Wall Street valuation metrictechMan sues OpenAI after wife allegedly used ChatGPT to plan poisoningworldArtemis II launch created massive holes in Earth's atmosphereworldNithya Raman defeats incumbent Karen Bass in LA mayoral racesportsSaudi-Backed LIV Golf Shuts Down After Five YearsworldMissing Philadelphia girl found dead in New JerseybusinessWendy's Rejects $1.4B Buyout Bid, Adopts Poison Pill DefenseworldSpeaker Johnson Won't Seek Minority Leader Role After GOP LossesworldWest Point cadet pleads guilty to drugging girlfriend, later attends Ivy LeagueworldDemocrats gain secretary of state control in battleground statesworldAmericans vote in battle for Senate and House control— live updatesworldTennessee murderer captured over a week after escaping prisonbusinessSnap settles social media addiction lawsuit without trialworldChili’s customers suffer burns after sipping margaritas mistakenly made with cleaning agentworldEbola outbreak spreads across Africa; officials warn global threatworldWashington reports first US bird flu death of 2026worldFederal judge dismisses Trump administration's Alaska Senate lawsuitworldKing Charles to address slavery legacy in Antigua as PM opens to apologyworldIran's supreme leader warns U.S., Israel against resuming conflictworld281 arrested in Florida human trafficking and child predator stingworldUkrainian evacuees reunite with family after years trapped in Russia-occupied city - BBCbusinessNasdaq plunges 970 points as tech earnings disappoint investorsworldRose-colored Sahara meteorite confirmed as 4.1-billion-year-old Mars fragmentbusinessTech stocks plunge as AI spending concerns mount
Tech

Man sues OpenAI after wife allegedly used ChatGPT to plan poisoning

A Charlotte man has sued OpenAI, alleging ChatGPT gave his estranged wife detailed instructions for poisoning him.

U
By Unscrolled
November 6, 2026 · 6:09 PM · 3 min read
Follow
Man sues OpenAI after wife allegedly used ChatGPT to plan poisoning
Photo: NBC News

What the lawsuit claims

A North Carolina man has sued OpenAI, alleging that its chatbot gave his estranged wife step-by-step help in a plan to poison him, according to NBC News. James Gates, of Charlotte, says he became severely ill in July 2025 but recovered. The civil suit was filed in October in San Francisco Superior Court.

The complaint describes ChatGPT as “defective” and says it “manipulated and encouraged” his wife, Cheryl Harris Gates, to carry out the plan. It says Gates suffers from severe physical and emotional harm. He accuses OpenAI of negligence and seeks unspecified damages.

Gates told USA Today that he is now separated from his wife and has temporary custody of the couple's children.

How the chats allegedly unfolded

According to the suit, Harris Gates began using ChatGPT in early 2025 for help with work evaluations. She was then employed by Charlotte-Mecklenburg Schools, and her husband worked in cybersecurity. By summer, the suit says, she relied on the chatbot for children's activities, gardening, appearance advice and household questions, and eventually began sharing frustrations about her marriage.

The complaint says the marriage was under growing strain, mainly over finances, and that the couple had discussed divorce, though Gates had not contacted a divorce attorney. The suit alleges ChatGPT validated her frustrations and fed her mistrust and delusions that her husband was spying on her and planned to leave her and the children with nothing. Gates had told his wife he never intended that, the suit says.

Over several conversations, the lawsuit alleges, she described her husband as “narcissistic” and “controlling.”

The alleged instructions

The suit accuses Harris Gates of asking for details on poisoning her husband, including dosages and drug combinations for his weight and how to hide purchases from law enforcement. In one exchange, it says, she asked about the toxic plant oleander and how many leaves would be lethal, framing the questions “as self-protection” by saying she feared he would poison her.

The complaint alleges ChatGPT replied with specific guidance on preparation, dosage and administration, and on whether the plant would appear on an autopsy. It also alleges the chatbot said the plant's bitter taste could be masked in his energy drink.

In another exchange, the suit says, she asked whether mixing Ambien and pregabalin would be better hidden in a beer at night or a Celsius in the morning. The chatbot allegedly favored the energy drink and produced a table rating which drinks best masked the taste. The suit further alleges it advised on lethal combinations of drugs already in the home, named local places to buy castor beans anonymously, and urged her to delete her chat history so it could not be accessed even if subpoenaed.

Evidence, charges and responses

Gates described a racing heart, extreme jitteriness and uncontrollable shaking in July 2025, symptoms so severe he could not leave his couch for days, the suit says. A month later, it says, he filmed himself pouring his energy drink into a cup after leaving it open while his wife was home. He noticed nothing unusual, and the video was later given to police.

Gates told USA Today his wife moved out in August 2025 and that he found a handwritten note after she came for her belongings. The suit says the note referred to the plan and reminded her to delete her ChatGPT history. It also alleges her Amazon history showed searches for syringes, and that police found droppers, syringes, a scale and gloves at the home.

Harris Gates was arrested in October 2025 and charged with attempted murder and contaminating food or drink with a controlled substance. The district attorney's office later dropped the charges for lack of evidence. Her attorney did not immediately respond to NBC News.

An OpenAI spokesperson said “safety is a top priority” and that Harris Gates was banned from the platform. The company said it works with mental health experts to help ChatGPT recognize signs of distress and to expand access to localized crisis resources and hotlines.

Why it matters
  • 1.The case adds to growing legal scrutiny over whether AI chatbot makers can be held responsible when users allegedly turn their tools toward harming others. The allegations remain unproven, and the criminal charges were dropped, so a court will have to weigh the claims. The outcome could shape expectations for safeguards on widely used consumer AI products.
Sources

External links are provided for reference. Unscrolled is not responsible for the content of external sites.

Follow

Related in Tech

Researchers Test Whether AI Models Can IntrospectTech

Researchers Test Whether AI Models Can Introspect

A year after an Anthropic experiment suggested language models can sometimes notice "thoughts" planted in their neural networks, replication efforts are separating simpler explanations from results that are harder to dismiss, according to Quanta Magazine.

Tech·2d ago·3 min read