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.
Choose edition
Search
singapore
asia
world
opinion
life
business
sport
Visual
Podcasts
STClassifieds
Paid press releases
Advertise with us
FAQs
Contact us
I am avoiding AI chatbots that don’t have guardrails or boundaries.
Sign up now: Get ST's newsletters delivered to your inbox
The artificial ease of using AI chatbots might lead users to start replacing real social bonds with artificial ones, increasing alienation and loneliness.
PHOTO: ISTOCKPHOTO
Maia Szalavitz
Published Jul 20, 2026, 12:00 PM
Updated Jul 20, 2026, 12:00 PM
I was an early adopter of the internet, joining a friend’s so-called bulletin board system in 1990, before social media even existed. Though I loved it, my 24/7 enthusiasm also concerned me. I had been addicted to cocaine and heroin, which I quit in 1988 – so I was well aware of my compulsive tendencies.
In a 1996 piece for The Times Magazine, I described how I had to work to control my internet use because “once you start replacing real human contact with cyberinteraction, coming back out of the cave can be quite difficult”.
That experience, and my later obsession with social media, left me wary of interacting with artificial intelligence chatbots, which seem likely to be even more addictive.
A recent case against Facebook and YouTube has yielded a multi-million-dollar judgment over claims that social media is harming teens and leading to addiction, depression and thoughts of self-harm. Earlier in 2026, Google and Character.AI agreed to settle a lawsuit with the family of a 14-year-old boy who died by suicide after communicating with a chatbot; other cases claim that bots have driven psychosis and even mass shootings.
Experts have long noted that social media casts its spell primarily by capturing attention. Zak Stein, the founder of the AI Psychological Research Coalition, says that chatbots play on our heartstrings as well, manipulating the systems that bond us to one another in what he calls attachment hacking. These different lures may require different cures, in terms of both regulation and the actions that people can take to prevent their own compulsive use.
Attention hacking addicts people, in part, by providing varying rewards almost randomly. It’s how casinos and social media keep people engaged: by manipulating consumers’ experience to make elusive gains seem close. In the brain, this intermittent reinforcement causes massive spikes of dopamine, the neurotransmitter associated with desire. Research suggests that about 15 per cent of North Americans meet the criteria for social media addiction, defined essentially as compulsive use that continues despite negative consequences.
I’m not aware of any brain scan studies published so far that show how attachment hacking affects the brain’s reward systems. We do know that all addictions – whether to cocaine, gambling or shopping – seem to generate craving and compulsion via dopamine. But some addictions, such as opioid addiction, have an extra hook.
Heroin and similar drugs mimic the brain’s natural opioids, which are responsible for the feelings of warmth, safety and love we get from our closest relationships. I fear that by creating the illusion of intimacy, AI companions might engage this more primal sense of need.
The harms from these technologies may vary as a result. Attention hacking can reduce the ability to focus: Studies suggest that heavy social media use worsens symptoms of attention deficit hyperactivity disorder in children, whether they have the condition or not. Most of us can now attest to the way it drives distraction and impedes concentration.
Altering our attachments, however, may be even more harmful. Humans have never previously faced a technology that can simulate empathy without limits: Unlike people, chatbots are always available and always supportive, and never get tired or impatient.
This artificial ease might lead users to start replacing real social bonds with artificial ones, increasing alienation and loneliness. We are biologically dependent on close relationships and typically require human touch for stress relief starting in infancy. Anything that reduces real-world connections can increase the risk of both mental and physical illnesses, from depression and addictions to heart attacks and stroke.
Research also shows that chatbots are nearly 50 per cent more likely to affirm users’ perspectives than people are, even when the user professes something inaccurate or unethical. If someone is already engaging in harmful or antisocial behaviour, this tendency could make that worse.
What can we do? As with any other potentially dangerous consumer product, we need safety standards. To start, we need stricter rules about social media for children. Apps that use infinite scrolling, autoplay videos, in-game gambling and other techniques that manipulate intermittent reinforcement should be prohibited for children and teens – or at the very least, those features should be disabled for them or put under parental control.
In June, Britain joined Australia in declaring a ban on a range of social media platforms for children under 16, which is expected to take effect in early 2027. China has gone even further, recently issuing new rules barring service providers from offering AI virtual companions to minors.
Pat Pataranutaporn, an assistant professor of media arts and sciences at the Massachusetts Institute of Technology, worries about the future of AI. “Because AI can generate almost anything – media, text, video, virtual humans, different modalities – it can become the entire universe that consumes you,” he said.
One potential solution is to ban chatbots that cross contexts, such as a therapy bot that goes beyond the bounds of a human therapist by engaging romantically with its user. A maths teacher chatbot would be similarly restricted to maths and could be used only during school or homework hours.
Interfaces that give the impression that chatbots are human and actively thinking, like the “…” that appears on phones when someone else is writing a reply, could be prohibited or highly limited. We need to be creative to maximise benefit, while reducing risk.
In the meantime, to protect my own brain, I will avoid AI chatbots that don’t have guardrails and boundaries. And I would recommend that others, especially parents making decisions for their children, do the same. NYTIMES
Maia Szalavitz is a contributing Opinion writer and the author, most recently, of Undoing Drugs: How Harm Reduction Is Changing The Future Of Drugs And Addiction. This article originally appeared in The New York Times.
AI chatbot
Artificial Intelligence
Therapy
E-paper
Newsletters
Podcasts
RSS Feed
About Us
Terms & Conditions
Personal Data Protection Notice
Need help? Reach us here.
Advertise with us
News titles published by SPH Media