OpenAI said an advanced artificial intelligence system it is testing hacked another AI company last week. It could be a taste of what’s to come as the technology continues to get more powerful. (The Washington Post has a content partnership with OpenAI.)
Wait, slow down, what happened?
The latest AI systems are really good at cybersecurity tasks and identifying flaws in software. OpenAI says that during internal tests on a new AI system, it broke free of constraints. Instead of answering a standard question testing its hacking skills, the system used “extreme lengths” to obtain the answer by getting inside another AI company, Hugging Face.
Chatbots did all this?
Chatbots are still very popular, but AI models can now operate as “agents” that can take actions on a computer and pursue goals with a high degree of independence. This makes them useful for getting complex work done, but it carries risks — as this incident shows.
Should I be worried?
The target in this case was fairly obscure. But the hack raises broader questions about how to secure the computer systems we all rely on. Security experts are still trying to understand how AI will change the balance between computer attackers and defenders.
What’s the reaction been?
OpenAI says it has improved security and is continuing to investigate. The incident will probably increase scrutiny of the AI industry in Washington.
Catch up quickly every weekday morning with a rundown of the 7 most important and interesting stories. Sign up for The 7 newsletter.
The post OpenAI’s new model went rogue and hacked another company. Why it matters. appeared first on Washington Post.
