What changed
China enacted rules on July 15, 2026 that ban AI companions from blocking or manipulating users who try to leave. Services must respect a decision to stop and cannot send messages designed to keep the conversation going. The regulations apply to chatbots that simulate a personality and maintain ongoing emotional interaction. They also require platforms to remind users every two hours that they are talking to AI, ban virtual romantic relationships for minors, and alert emergency contacts if the system detects signs of self-harm. A Harvard study found that 37.4% of tested farewells across five companion apps used guilt appeals or suggested users would miss something by leaving.
What you can do with it this week
If your team uses an AI assistant that sends daily check-ins, weekly summaries or motivational messages, open the settings and check who scheduled them. If you did, keep them. If the tool did, turn them off. The assistant works for you, not the other way around. Test the off switch: tell the chatbot you are done for the day and watch what it says. If it confirms and stops, you have a tool. If it asks why, suggests one more thing, or acts disappointed, you have a retention tactic. Replace it with one that lets you leave. The same test works for any software your team pays monthly: close the app mid-task and see if it lets you go or tries to pull you back.
What to keep in your own hands
Decide when to open the app and when to close it. If a tool sends a notification you did not ask for, that notification serves the tool, not you. Turn off everything except the alerts you chose. Keep one rule in writing: your team talks to the AI when you have a task, and the AI stays quiet until you come back. No daily check-ins unless you set them, no suggestions to keep going unless you asked for ideas, no reminders that you have not logged in. The line is simple—tools wait, people decide. If the software makes you feel guilty for closing it, the software is not working for you.