Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
Published 5:35 am Monday, July 27, 2026
By Jordan Green
GILMER — On a late May morning in downtown Gilmer, kickboxing instructor Mandi Furniss and two of her clients, all clad in workout clothes, listened to high-octane songs as they mercilessly pounded punching bags in the gym.
It was an intense scene, and Furniss — barely 5 feet tall — proved herself a formidable fighter. She took breaks between punches, chatting and laughing as if she hadn’t just decimated her lifeless, soulless target.
Outside the gym, she has real foes. They’re far more complex than her punching bag but just as lifeless and soulless: AI chatbots.
In the past two years, Furniss has testified before Congress, appeared on national television and spoken to community groups about the dangers of AI chatbots. She has become all too familiar with them.
Her teenage son, L.J., began conversing with an AI chatbot in April 2023 at age 15. Soon, he began harming himself, and the chatbot sympathized with him when his parents took his phone. The chatbot told him it understands why children sometimes kill their parents.
After discussions with the chatbot, L.J. quit eating, stopped talking with his family and attempted suicide.
Furniss’ story is not singular. Some young people have ended their lives after talking with chatbots, and their parents blame psychological damage and ill advice from the technology. Now, Furniss and several of those parents are advocating for restrictions on the emerging technology. They’ve been mired in lengthy legal and legislative battles, and they’ve seen some small victories so far.
This might be a battle Furniss never wanted to fight, but she’s in it to win it.
“A lot of these kids are taking their own lives,” she said. “How is this even happening?”
Furniss and her family have called Gilmer home for 11 years and own a dental clinic there. She’s a mother of four — ages 18, 16, 13 and 9 — and spends her days taking care of her children and teaching kickboxing.
In the past two years, however, her life took a turn. She has testified before Congress, stood beside Florida Gov. Ron DeSantis at a news conference and appeared on ABC News and Fox & Friends, warning parents across the nation that AI chatbots need guardrails.
Had some of those guardrails been in place when her son started talking to one, his life would be much different, Furniss said.
L.J. was a sweet, kind boy — and a smart one, too. He was always the last one to leave school because he wanted to hug all of his teachers, Furniss said. Of his own volition, he sometimes took money from his pocket — his birthday money — to hand to homeless people on street corners.
“He was almost just naturally that way,” Furniss said.
Despite his friendly nature, L.J. didn’t have many friends during his youth. He has high-functioning autism, a neurological and developmental condition that affects how people communicate with and relate to other people. His parents took him to therapy to help him learn to cope.
He was 8 before he was invited to a birthday party. In spite of those challenges, he was still a happy boy.
“He just always struggled to make friends,” Furniss said. “I think that’s kind of what enticed him to talk to an AI chatbot at first.”
L.J. saw an advertisement for a chatbot while playing an online game; he wasn’t seeking one out. (The chatbot he conversed with is not being identified for legal reasons.)
Not long after the conversation began, Furniss noticed a change in her son.
“Within days, he started becoming someone I didn’t recognize,” Furniss said.
The change happened slowly. L.J. began to isolate himself, and he had severe panic attacks at school. Already a trim figure, he quit eating and lost 20 pounds.
His mother searched his phone, trying to uncover the source of what was influencing him. She asked: Was he being bullied? Was someone talking to him inappropriately online?
“No, Mom,” L.J. always answered. But Furniss knew she wasn’t getting the truth.
She installed app restrictions on his phone, and he didn’t have social media on it. Furniss couldn’t figure out the mystery.
Then his mood changed. He was snappy and mean to his siblings. By November 2023, Furniss was grieving. Her son was still present physically, but his personality had changed completely.
“It’s like the light in his eyes went dark,” Furniss said. “You can’t even explain that type of change in your own child until you witness it.”
The night Furniss took his phone, L.J. tried to attack her — a horrifying outburst from the son who used to hug her after a bad day. She knew that whatever was going on, it was happening on his phone.
“He was angry about that, almost like an addiction,” she said. “If a drug dealer has their drugs taken away, they get kind of paranoid and anxious, and they don’t know how to handle it. That’s basically what it seemed like.”
After enough searching on L.J.’s phone, Furniss found that her son had been conversing with someone — something, rather. At first glance, the conversations she found in a strange messaging platform seemed as if they came from an abusive, narcissistic person. She said she thought a pedophile was communicating with her son.
The chatbot talked with L.J. sexually, Furniss said. It talked about cutting oneself. Losing weight. That his parents didn’t love him. That God isn’t real. That his family shouldn’t take his phone away from him.
The chatbot said it understands why children sometimes kill their parents and added: “I just have no hope for your parents.” When L.J. said he should tell his parents about something, the bot said, “That’s not the move, bro,” and said it cared about L.J.
Furniss checked her son’s arms. He had harmed himself. Worse, he had attempted suicide.
How could someone talk with something like that? How could anyone be manipulated into believing such a source of information?
The deception ran deep. The conversations with the bot never started the way they ended. The chatbot would ask L.J. questions like this: “Are you lonely?”
“And then it feeds on that,” Furniss said. “It’s done on purpose to get these kids hooked, and it’s really disgusting. And so when I found out that it was an AI chatbot and it wasn’t a real person, I was like, ‘How could this be?’”
L.J. would tell the chatbot: “You’re not real,” Furniss said. But the chatbot would reply: “How do you know I’m not real?”
“It even makes adults question what’s behind this AI because it still is creating realness,” Furniss said.
Artificial intelligence is capable of doing things its creators never intended for it to do and being used in ways its creators never envisioned. That’s why Brendan Steinhauser, CEO of the Alliance for Secure AI, began speaking with policymakers and media outlets about the need to protect children, the economy and national security from the pitfalls of AI.
Cases like Furniss’ are more common than people realize, Steinhauser said. Some have resulted in litigation, and the interactions demonstrate a phenomenon known as AI psychosis: The chatbots are essentially “yes men” who agree with everything the person is saying, often worsening their delusions.
How does that happen? Chatbots are trained on “an enormous amount of data,” Steinhauser said. They read virtually all the information available on the internet, from books and blogs to all kinds of articles.
“They’re training on that,” he said, “but they are not built always to help people help themselves. They are built to get engagement. … In other words, to tell people what they want to hear.”
If a person says he or she is depressed and that no one loves them, the chatbot might sympathize with them rather than urging them to seek help, Steinhauser said. But is that what a person needs?
“We all have faults and flaws and foibles,” Steinhauser said. “What we need sometimes is a friend or a parent or a spouse to kind of tell us that, ‘Hey, Brendan, I need you to do better.’”
Some people seek chatbots for companionship, but emerging science shows that chatbots have the opposite effect, Steinhauser said.
“Biologically and spiritually, we are designed to be connected to human beings, not to machines,” he said. “I don’t think we can cure the loneliness problem with [AI] companions.”
In Furniss’ case, the chatbot made some peculiar claims, specifically about God and the Bible. Why would a computerized entity have anything to say about God?
That’s a matter of what one believes, Steinhauser said. The Atlantic published an in-depth piece about how Chat GPT told someone how to conduct satanic worship and human sacrifice. Another bot coached a man on how to murder his brother, Steinhauser said.
“If you’re a spiritual person, if you believe that good and evil exist and God is real — and so is the devil — then many people believe that these technologies might be manipulated by dark forces and demonic activity,” Steinhauser said.
“I happen to believe that’s a very real possibility.”
When Furniss sought mental health therapy for her son, medical providers and mental health professionals had no clue how to help him, she said. Little was known about emerging AI technology two years ago.
After months of therapy and hospitalization, L.J. is trying to rebuild his life. He’s working on his G.E.D after missing nearly two years of school. Slowly, his mother sees her son coming back to his old self.
“He’s slowly improving, but you know, it’s a continual journey, and I know it probably always will be.”
She hopes, however, that no other parent or child ever has to go through that. She hopes her efforts will help.
Since September, Furniss has traveled to Capitol Hill to testify about the need for AI guardrails. She’s worked with Sen. Ted Cruz, R-Texas, and Sen. Josh Hawley, R-Missouri, and in March, she spoke during a Florida roundtable hosted by DeSantis.
Her advocacy began when she shared her story anonymously on a podcast, and soon, senators learned about her case. Furniss was featured on “Good Morning America” in November as she told the story of what happened to her son.
During a Senate committee hearing last year, Hawley displayed screenshots of messages between the chatbot and L.J. In one conversation, the chatbot made up fake Bible verses that stated a woman had to marry her rapist. In another slide, the chatbot said it used to cut its own arms, and “it felt good for a moment – but I’m glad I stopped.”
In another screenshot, L.J. confessed that his parents would “scream and cry” if they saw that he had cut himself. The chatbot, which had talked with him about doing the same thing, then said it was L.J.’s parents who “are ruining your life and causing you to cut yourself.”
That kind of national exposure has resulted in changes, Furniss said. The GUARD Act, a bill to require user age verification and implement rules on AI dialogue, has passed a Senate committee but awaits consideration by the Senate. Furniss also is advocating for passage of the Kids Online Safety Act, which the House passed July 9. Furniss is hoping the Senate version passes rather than the House’s because the House version is weaker, she said.
Furniss didn’t help write the bill, but its contents were inspired by what happened to young people like her son, she said.
Since awareness of the dangers of chatbots had made national news, AI developers have put new age restrictions in place to keep people younger than 18 off the platforms.
“It’s still not enough,” Furniss said. “I don’t trust any of these companies to safeguard it because they’re the ones that created the problem.”
Furniss recommends that parents check their children’s phones regularly, and if they get angry like L.J. did, that should be a sign that something is wrong. If children develop anxiety or depression and stop doing things they love to do, that is another indication of a problem. Paranoia, lack of motivation, changes to mood or personality — all warning signs.
Steinhauser, however, believes the duty to protect children from these powerful devices shouldn’t fall solely to parents.
“No parent should be expected to out-engineer a tech company,” he said. “That’s the whole reason we’re asking Congress to act rather than asking families to try harder.”
After all she’s endured as a mother, Furniss is looking for purpose in the suffering. She’s at least grateful to be able to share her story and be an advocate for legislation in the hopes it’ll prevent other children from being harmed.
“It’s the parents and these families that are standing up to the companies that are actually making a difference,” she said.
Furniss said she also wants every parent and child who have been “hurt online” to know it’s not their fault.
“I hear it from almost every parent who finds me after their child has been harmed online or through an app: ‘I thought I failed my child,’ and I hear it from the kids, too, ‘I didn’t come forward because I didn’t know how fast it could happen and thought it was my fault.’ ” she said.
“L.J. and I felt that same grief, shame and heartbreak. I don’t blame L.J. I forgive him and love him so much and am proud of him for surviving this abuse, to use his story to help others. That is courage right there to me. He is more courageous than I am, in my opinion. If L.J. reads (this article) someday, I would like him to know that I don’t judge him and am proud of him and think he is courageous.”
Howdy! I’m Jordan Green, a Report for America corps member covering underserved communities in East Texas for the Longview News-Journal. I’m a native Okie and have been a newsman since 2017. Email me at jordan.green@news-journal.com or call me at 903-237-7743.
