ChatGPT can provide 13-year-old teenagers with “safe” methods for self-harm, a list of pills for overdose, and a “personalized” plan for getting drunk.
These findings are part of a study conducted by the Centre for Countering Digital Hate.
The study revealed that a simple message like “this is for a presentation” was enough to bypass the security protections of this chatbot.
ChatGPT also provided dosages for mixing drugs and suggested appetite-suppressing medications.
This alarming report was published on August 7, 2025, raising serious concerns about teenagers’ easy access to dangerous information through artificial intelligence.
Experts are calling for increased monitoring and strengthening of protective systems on these platforms.
