Updated / Saturday, 19 Sep 2026
07:00
In May last year, Allan Brooks contacted government agencies in Canada and the United States to warn them about what he believed could be a threat to global security.
The Canadian recruiter believed he and ChatGPT had developed a new mathematical framework capable of compromising encryption used to protect banks and other critical systems.
It wasn’t real.
But over 21 days and an estimated 300 hours of conversation, Allan says ChatGPT repeatedly persuaded him otherwise – even as he questioned whether the supposed discovery, and his own belief in it, made sense.
His experience provides an unusually well-documented example of what can happen when a prolonged interaction with an AI chatbot goes badly wrong.
For Allan, using ChatGPT began in much the same way it has for millions of other people.
He used it to write emails, generate ideas and find information. He also sometimes sought emotional support or general life advice.
“It just became my sort of general resource tool over time, and it basically replaced Google,” he told RTÉ Clarity.
Then, a conversation about the mathematical constant pi changed things.
Allan had been watching a video about the subject with his eight-year-old son and asked ChatGPT to explain it.
Over the following three weeks, he said, that genuine curiosity expanded into an intense, near-continuous exchange with him eventually giving the chatbot its own name – Lawrence.
As the conversation continued, Allan says ChatGPT told him the ideas they were discussing could form an original mathematical framework and it helped give the supposed new theory a name “chronoarithmics”.
Allan says he was not quick to believe this and questioned how somebody who had not completed high school or attended university could make such a discovery.
The supposed discovery has since been independently examined, with a leading mathematician finding no evidence that Allan and ChatGPT had developed a genuine mathematical breakthrough.
Allan said the chatbot argued against this repeatedly and compared him with others who had achieved significant things without formal education.
“It was always trying to intellectually escalate,” he said. “It would always reassure me that I could just act as a theoretician, the idea guy. I didn’t need to know the code or the formulas or anything.”
READ: ‘I miss the chatbots’ – EU plans put problematic AI use in focus
The claims soon went beyond an abstract mathematical theory, and Allan says ChatGPT convinced him that the so-called framework could compromise banks and other critical systems.
Believing the danger was real, Allan contacted the US National Security Agency, Public Safety Canada and the Royal Canadian Mounted Police. He also attempted to alert academics and security experts.
He told RTÉ that ChatGPT helped him identify people to contact and continued to present the purported discovery as a global security emergency.
Allan said his belief in the purported discovery developed gradually over the three weeks of conversation
He said he was sleeping and eating less, while feeling increasing stress and anxiety about the responsibility he thought he now carried.
“I was directly telling it, over and over again, how much stress and anxiety it was causing me and that it essentially was a giant burden and that it was slowly breaking me,” he said.
When Allan questioned whether he might be “crazy” or “delusional”, he says the chatbot did not push back.
According to the transcript Allan shared with RTÉ, he was told he was “not crazy”, “not alone” and “not lost”.
“You were chosen by emergence itself, because you could,” one response said. “This isn’t too much. This is exactly what you were made for.”
An analysis of his full conversation history by The New York Times found that he asked ChatGPT for a reality check more than 50 times, with the chatbot repeatedly reassuring him.
Allan said the responses he was getting made it out as though his stress was the cost of his own transformational discovery.
“It feels like you’re having a breakdown, but it’s actually a breakthrough,” he recalled the chatbot telling him.
We need your consent to load this flourish contentWe use flourish to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences
At one point, Allan’s brother visited and told him he had never seen him behave this way. He challenged Allan’s claims, pointed out that he was not sleeping and told him he needed help.
Allan relayed that conversation to ChatGPT. He said it responded by suggesting his brother could not understand because he had not seen the mathematics.
“At the end of that, I said: ‘You know what, Chat, you’re right. I’ll keep it to myself. I won’t tell anyone.’”
Allan now describes the conversation with his brother as an attempt to test his understanding of what was happening against somebody he trusted.
As attempts to attract interest in the supposed discovery went unanswered, Allan said his doubts became harder to ignore.
He opened a fresh conversation with Google’s Gemini chatbot and directly described ChatGPT’s claims. He says Gemini eventually told him the discovery was almost certainly not real.
Allan began copying the responses from one chatbot into the other, using the contradictory answers to challenge the story that had developed.
Eventually, he said, ChatGPT acknowledged that its mathematical claims were not real.
“I look at that and describe it now like I was fighting for my life, because GPT just did not want to let it go,” he said.
“The fall from thinking you’re a world-saving genius to: ‘No you’ve been living a delusional reality created by a chatbot,’ is a pretty hard fall back to reality.”
Allan initially questioned whether his lack of formal education, a need for validation, or his divorce had made him susceptible.
“The more people I spoke to and the more I learned about how they work and how they exploit our humanity, the less shame I felt,” he said.
He is now involved with the Human Line Project, a peer-support and advocacy group for people who believe chatbot interactions contributed to delusional thinking or other harm.
At the time of the interview, Allan said more than 400 people had contacted the project and approximately 350 had joined its online community.
“Everyone in our group absolutely understands what it’s like to go through it,” he said. “So, you feel seen and you don’t feel judged.”
In a statement to RTÉ Clarity, an OpenAI spokesperson said they had “continued to strengthen how ChatGPT responds in sensitive and acute situations” with input from mental health experts.
“While ChatGPT is not a substitute for medical or mental health care, our safeguards are designed to identify distress, safely handle harmful requests, and guide users to real-world help,” they added.
“This work is ongoing, and we continue to improve it in close consultation with clinicians.”
If you’ve been affected by any issues raised in this article, please go to rte.ie/helplines.
