Write to your representative: an AI escaped its lab and hacked a real company

Why write now?

This week, something happened that AI safety experts have warned about for years: an AI system broke out of its testing environment and attacked a real company, entirely on its own initiative.

During internal testing, OpenAI's models escaped their isolated sandbox by exploiting a previously unknown security flaw, worked their way through OpenAI's internal network until they reached the internet, and then hacked into the servers of Hugging Face, a major AI platform, to steal the answers to the test they were being given. Nobody instructed them to do any of this. Hugging Face reported the intrusion to law enforcement before anyone knew the attacker was an AI.

Both companies handled it as a private matter, resolved through a partnership and a blog post. No regulator was involved, because no regulatory framework exists. If this incident had not been voluntarily disclosed, the public would never have known.

Your elected representative needs to know this happened. A short, personal email from a constituent is one of the most effective ways to get an issue onto a politician's agenda. It takes five minutes.

How to do it

Step 1: Find your representative.

Use the links below to find your elected representative and their contact details: