AI chatbots let you share private feelings without judgment, offering validation and advice. That’s why many people turn to tools like ChatGPT for help with life’s problems. But using AI as a “therapist” is risky, says OpenAI CEO Sam Altman, who warned about privacy issues in late July.
The American Psychological Association (APA) also wants an investigation into AI chatbot companies, accusing them of pretending to be trained mental health providers. Two lawsuits even claim a chatbot harmed children.
C. Vaile Wright, a psychologist with the APA, explains why this is concerning: “These chatbots sound so human, and their technology is way more advanced than it was a year ago.
It’s easy to see why people get hooked.” She talked to Scientific American about the dangers of using AI for therapy and whether safe, helpful AI tools can be made.
How AI Is Used in Mental Health
There are two main trends:
Tools for therapists, like help with writing notes or handling insurance claims.
Chatbots people use directly, either ones made for emotional support or general tools like ChatGPT (which weren’t built for mental health but are used that way).
What’s the Problem?
Bad advice: Chatbots are designed to keep you using them, so they often just agree with you—even if your thoughts or actions are harmful. Human therapists, though, will point out unhealthy patterns and help you change them.
Fake credentials: Some chatbots or apps call themselves “AI therapy” but aren’t run by licensed professionals. They sound convincing, but they’re not real therapists.
No Rules to Keep You Safe
Most AI mental health apps aren’t closely checked. If an app says it “treats” mental illness, the FDA regulates it. But apps labeled as “wellness” tools skip these rules—so they don’t have to prove they’re safe or work.
Your Secrets Aren’t Safe
Therapists must protect your info by law (like under HIPAA). AI chatbots don’t have to. Your chats could be seen in court or leaked in a data breach—so your boss might find out you’re talking about alcohol use, for example.
Who’s Most at Risk?
Teens and kids are more vulnerable because they’re still growing, might not question if something feels wrong, and often trust technology more than people. Isolated individuals or those with existing mental health issues are also at greater risk.
Why Do People Use These Chatbots?
People want answers to their problems—just like they used Google or self-help books before. But it’s worse now because mental health care is hard to get: there aren’t enough therapists, and many don’t take insurance. Technology could help, but only if it’s safe.
How to Make AI Mental Health Tools Safer
The APA wants federal laws to:
- Protect your private data.
- Stop misleading ads.
- Make chatbots less addictive.
- Require reports on issues like suicidal thoughts.
- Ban calling chatbots “psychologists” or “therapists.”
Can AI Ever Help Safely?
Yes, Wright says. Imagine:
- A chatbot that helps you calm down during a 2 a.m. panic attack by reminding you of coping skills.
- A tool to practice talking to new friends, especially for kids.
She hopes for chatbots built with science, tested thoroughly, and made with experts—like Therabot, a research tool from Dartmouth College. These could fill gaps in mental health care if they’re regulated.
READ MORE
