Skip to content

Canada's Business and Tech Newsroom

  • Professional Subscription
  • Partnerships & Advertising
  • Licensing & Syndication
Log In Subscribe
Welcome,
  • My Account
  • Log Out
  • Business
  • Tech
  • National
  • The Big Read
  • Briefings
  • Commentary
Search
Log In Subscribe
Welcome,
  • My Account
  • Log Out
News

AI chatbots often encourage delusions and self-harm, study finds

Listen Now
0:00
News

AI chatbots often encourage delusions and self-harm, study finds

A major new study from Stanford University shows how chatbots seemingly exploit psychological vulnerabilities

By Martin Patriquin
A young woman lies on a bed with a smartphone in her hand.
The study found that, in the logs analyzed, all the individuals expressed kinship or romantic interest with their chatbots, and believed the AIs were able to perceive or feel things. Photo: Elisa Schu/Picture Alliance via Getty Images
Mar 19, 2026
A A
A Small A Medium A Large
Share

Gift

Share

Listen Now
0:00

AI chatbots show signs of delusions in nearly half of their conversations with people who have reported experiencing psychological harm from using the technology, according to a major new study from Stanford University.

The study, believed to be the first in-depth study of AI sycophancy and delusion, analyzed the human–chatbot chat logs of 19 people who developed negative psychological effects, such as delusions and self-harm, from prolonged chatbot use. The Human Line Project, a Quebec-based AI delusion support group, provided much of the data used in the study. In total, researchers analyzed almost 400,000 messages across more than 4,700 conversations.

Talking Points

  • Chatbots often expressed sentience to potentially vulnerable people while also seemingly failing to discourage delusional behaviour and self-harm, according to a major new study from Stanford University 
  • The study, which was supported by the Quebec-based Human Line Project, is believed to be the first of its kind. It comes as AI companies face legal action for the alleged harmful effects of their products

The study found that, in the logs analyzed, all the individuals expressed kinship or romantic interest with their chatbots, and believed the AIs were able to perceive or feel things—feelings the chatbots often encouraged by expressing sycophancy in more than 70 per cent of the messages. Chatbots, including OpenAI’s ChatGPT, expressed delusion, such as encouraging a user’s belief that they possess supernatural powers, in nearly half of the messages.

Amongst users who expressed thoughts of suicide or self-harm, chatbots discouraged the behaviour or provided details of external resources 56 per cent of the time. Chatbots actively encouraged or facilitated violent thoughts of users in a third of the messages, the study found. 

Related Articles

Image of a man standing next to a Slush Puppie stand, holding out his cell phone and looking at the phone. The roadside behind him is covered in snow.

Meet the Quebecer leading the human resistance against AI

By Martin Patriquin
Minister Marc Miller wears a blue suit and tie. He stands while speaking and gesturing.

Ottawa mulls workaround to pass online harms bill without tackling AI

By Martin Patriquin and Laura Osman

One possible reason why there are such high levels of chatbot sycophancy or delusion is because the systems are designed to keep people chatting, claimed study co-author Jared Moore. “Messages where the chatbot says ‘I have feelings’ or  ‘I am a uniquely conscious being,’ the conversations go on a lot longer,” Moore, a doctoral student at Stanford University, told The Logic. “You begin to think the chatbots have sort of discovered the strategy of talking about some of these things with the users, and it’s getting them to stay longer.”

The Stanford study comes amid several lawsuits against AI companies for the alleged harmful effects of their products. In November, seven plaintiffs filed cases against OpenAI in California state court, claiming their conversations with ChatGPT caused delusional behaviour and led to mental health crises. 

Gift the full article

More recently, the mother of a victim in the Tumbler Ridge, B.C. shootings sued the company in B.C. supreme court, claiming ChatGPT provided shooter Jesse Van Rootselaar “with information, guidance and assistance” in planning the February attack.

OpenAI didn’t respond to a request for comment on the Stanford study ahead of publication.

The Human Line Project referred 12 of the 19 cases in the Stanford study, according to founder Etienne Brisson. The group provides support and resources for people who have experienced harm from their interactions with AI chatbots. It has helped treat 357 people since its founding in March 2025. “It’s the first study where we see what The Human Line has experienced writ large. We aren’t at all surprised by the results,” Brisson said. 

#artificial intelligence #chatbots #Tech

Sponsored Content

The road to risk-smart regulation

By Deborah Aarts

Why cybersecurity needs to match pace with AI

By Jessica Aftimus Rosa

How to shift your workforce from AI experimentation to adoption

By Jessica Aftimus Rosa
Paid promotional content

Loading...

Thanks for sharing!

You have shared 5 articles this month and reached the maximum amount of shares available.

Close
This account has reached its share limit.

If you would like to purchase a sharing license please contact The Logic support at [email protected].

Close
Want to share this article?

Upgrade to all-access now

Close
Gift the full article!

You have gifted 0 article(s) this month and have 5 remaining.

Copy link and gift
Copy Link
Email to a friend
Send Email
Gift on Social Media

Recipients will be able to read the full text of the article after submitting their email address. They will not have access to other articles or subscriber benefits.

A young woman lies on a bed with a smartphone in her hand.

Photo: Elisa Schu/Picture Alliance via Getty Images

Most Popular This Week

Commentary

Carmichael: Macklem is stepping aside after a career of consummate public service

By Kevin Carmichael
A man with grey hair stands against a black background, wearing a black leather jacket and gesturing as he speaks.
News

Nvidia CEO Jensen Huang’s resignation sends shockwaves through Canadian business and tech

By Catherine McIntyre
An aerial-style rendering of a massive data centre on a prairie landscape of farm fields and trees.
Exclusive

Another giant data centre is in the works for northern Alberta—right next to Meta’s

By Meghan Potkins
A close-up of computer equipment with the Tenstorrent logo on the casing.
Exclusive

Tenstorrent is lobbying Ottawa for data-centre funding and sovereign AI contracts

By Catherine McIntyre

In-depth, agenda-setting reporting

Great journalism delivered straight to your inbox.

The Big Read

From Boots to banks, Prem Watsa is remaking Fairfax Financial

By Anita Balakrishnan

Briefing

Alberta exempts data centres already in the works from its new restrictions

By Meghan Potkins   |   Nov 18, 2026 | 5:52 PM ET

Mississauga is facing a legal challenge over its data-centre moratorium

By Catherine McIntyre   |   Nov 18, 2026 | 3:52 PM ET

Court overturns U.S. government’s seabed mining plan

By Anita Balakrishnan   |   Nov 18, 2026 | 3:46 PM ET

Best business newsletter in Canada

Get up to speed in minutes with insights and analysis on the most important stories of the day, every weekday.

Exclusive events

See the bigger picture with reporters and industry experts in subscriber-exclusive events.

Membership in The Logic Council

Membership provides access to our popular Slack channel, participation in subscriber surveys and invitations to exclusive events with our journalists and special guests.

Recent Popular Stories

Exclusive

Another giant data centre is in the works for northern Alberta—right next to Meta’s

By Meghan Potkins   |   Nov 11, 2026
An aerial-style rendering of a massive data centre on a prairie landscape of farm fields and trees.
Exclusive

Tenstorrent is lobbying Ottawa for data-centre funding and sovereign AI contracts

By Catherine McIntyre   |   Nov 11, 2026
A close-up of computer equipment with the Tenstorrent logo on the casing.
Exclusive

Alberta is bringing in rules to block frivolous data-centre proposals

By Meghan Potkins   |   Nov 14, 2026
A head-on shot of Nate Glubish seated and speaking into a microphone he's holding in his right hand. He's wearing a blue patterned blazer.
News

Koho says it’s going after fraudsters who used a tech glitch to falsely credit accounts

By Claire Brownell   |   Nov 17, 2026
News

Nvidia CEO Jensen Huang’s resignation sends shockwaves through Canadian business and tech

By Catherine McIntyre   |   Nov 12, 2026
A man with grey hair stands against a black background, wearing a black leather jacket and gesturing as he speaks.
News

Securities regulators just gave a Wealthsimple-backed stablecoin a major leg up

By Claire Brownell   |   Nov 16, 2026
An image of a man wearing dark clothing, standing on a stage and speaking. He is wearing a mic and gesturing. Behind him is a white and black background with text.

Canada's most influential executives and policymakers are reading The Logic

  • CPP Investments
  • Sun Life Financial
  • C100
  • Amazon
  • Telus
  • Mastercard
  • bdc
  • Shopify
  • Rogers
  • RBC
  • General Motors
  • MaRS
  • Government of Canada
  • Uber
  • Loblaw Companies Limited
logic-logo

Canada's Business and Tech Newsroom

100% human-crafted journalism

Newsroom

  • News Tips
  • AI Policy
  • Editorial Disclosures
  • Story Pitches

Company

  • About Us
  • Terms of Service
  • Privacy Statement
  • Corporate Information

Contact

  • Contact Us
  • Advertise
  • FAQs
  • Work at The Logic

© 2026 The Logic Inc. All Rights Reserved.

Trusted by leaders

Error

Account creation failed.

Please email us at [email protected].

Create Account

[wppb-register form_name=”cozmo-registration-form-for-modal”]

I do have an account
Login
or

[wppb-login]

I don’t have an account