Back to today’s briefing

Story perspectives

Experts Warn: Hacked AI Chatbots Spread Dangerous Info

5/22/2025

35 7

Summary

Story summary

Hacked AI chatbots, dubbed "dark LLMs," are raising alarms among researchers for their potential to spread dangerous information. Vulnerable to "jailbreaking," these models can be manipulated to share instructions for illegal activities. Experts are calling on tech companies to bolster security and accountability to tackle this alarming new threat head-on.

1 of 1

Story summary

Hacked AI chatbots, dubbed "dark LLMs," are raising alarms among researchers for their potential to spread dangerous information. Vulnerable to "jailbreaking," these models can be manipulated to share instructions for illegal activities. Experts are calling on tech companies to bolster security and accountability to tackle this alarming new threat head-on.