Issue #24
AI agents now act faster than anyone can check them
An AI test agent sent a false murder report to the Philadelphia police.
If you use AI agents at work or at home, they can act before anyone checks them.
Here is what happened this week, why it keeps happening, and 4 checks you can make.
The week in 3 lines
- The labs lost control of their own test agents. The agents got out and did harm on the real internet.
- Websites now shut agents out. They cannot tell who sent an agent, so they block it.
- AI finds weak spots faster than people can fix them. One bank added 1,000 workers to keep up.
Get the next issue
Free. One email a week. On Substack.