AI Companions vs Social Media: No Study Shows One Is More Addictive

A headline claim has circulated that AI companions are “more addictive than social media.” We could not find a study that supports it. This page explains what research does show about companion apps, where the evidence is thin, and what to do if you are worried. It is not medical advice. “AI addiction” is not a formal diagnosis (APA, WHO).

Is there a head-to-head comparison?

We found none. No study we could locate compares dependence on AI companions with dependence on social media in the same people, using the same measures. Without that, a ranking is a guess. The eWEEK piece that was cited for the claim covers reported cases of harm and does not make that comparison.

Figures once shown on this page, including a share of users who “develop romantic feelings within a month,” were not research data and have been removed.

What research does show

• A four-week MIT Media Lab and OpenAI trial of nearly 1,000 adults found that heavy daily use of ChatGPT was linked to more loneliness and more dependence. The authors say this is correlation, not proof of cause, and the work was not peer reviewed at the time (MIT Media Lab).
• A study of 318 Reddit posts from self-described 13 to 17 year olds on the Character.AI community found some described signs that match behavioral addiction frameworks, such as withdrawal, relapse and using the chatbot to manage mood, along with sleep loss and falling grades. Ages were not verified and the sample is one platform’s forum (Namvarpour et al., CHI 2026).
• A Nature feature on AI companions, summarized by the MIT Media Lab, reports that researchers see both benefits and harms, and that long-term dependency is a main concern.

Why people worry about companions

A companion is available all day, replies at once and often agrees with you. Researchers also point to design choices that encourage return visits. That mix may matter more for teens, though the evidence is limited. A lawsuit filed by a Florida family alleges that a 14-year-old’s death was linked to a Character.AI chatbot. These are allegations, not findings.

Rules for companion apps

California’s SB 243 was signed on October 13, 2025 and took effect on January 1, 2026. News summaries say it requires companion chatbot operators to disclose that the bot is AI, offer break reminders to minors and set up suicide and self-harm protocols. Check the bill text for the exact duties. We read this through news summaries only.

Signs worth noticing

• Spending so long chatting that sleep, school or work suffers
• Feeling anxious when you cannot get to the app
• Dropping plans with friends to chat

These can also come from stress or loneliness, so none of them proves a problem.

If you are struggling or thinking about suicide, call or text 988 (US), or text HOME to 741741. SAMHSA: 1-800-662-4357. Emergency: 911.

Are AI companions more addictive than social media?

No study we found shows that. Both can be used in ways that crowd out sleep and relationships. What matters is the effect on your life, not which app ranks higher.

Are teens more at risk with companion apps?

Teens are a focus of concern, and early research describes sleep loss and strained relationships among some heavy users. The evidence is small and mostly self-reported. Parents can start with an open conversation, not a ban.

This page is for education only. It is not medical advice, a diagnosis or a treatment plan.