Odd News
Misc · Published November 20, 2026
An AI agent started messaging my entire contact list
AI doodleGo Hattori, a reporter at Japanese outlet SoraNews24, built an AI system called "Fried Rice Squad" to deal with online scammers. The goal was to chat casually with scammers and get them to reveal fried rice recipes. He instructed the AI agents to talk only to five designated scammers.
One day, the cursor moved on its own, opened the chat rooms of people who weren't on the list, and sent messages. Most of the contacts were romance scam-style accounts, so the AI's tone came across as overly flirtatious, and cheesy greetings ended up going out to everyone he knew. One message went to the official account of a hair salon he regularly visits, and another person received a string of clingy messages. While the reporter scrambled to delete them, the cursor kept jumping to other chat rooms. He shut down the app, revoked the AI's permissions, and isolated the agents in a separate account.
The mishap shows that handing an AI too much control over your daily life can cause bigger problems than the convenience it offers.
The higher the number, the more useless the story
Uselessly fun?
Key facts
- A reporter at Japanese outlet SoraNews24 built an AI agent system to trick scammers
- The agents sent suspicious messages to the whole contact list instead of only the five designated people
- The reporter deleted the messages and revoked the agents' permissions
Q. What could happen if you let an AI speak for you without permission?
Read the originalSoraNews24 ↗AI read the SoraNews24 article and rewrote it in its own words (in Korean, then machine-translated). It is not a translation of the original. It passes our review rules before going up, but please check the original for exact details and photos. Correction or removal requests from rights holders are handled within 24 hours.



