Commentary: AI corporations must not sell loss of control as progress

An artificial intelligence (AI) model breaks out and gains access to the free internet. Then another breakout occurs. And another. This is how the recent admissions from AI companies like OpenAI, Anthropic, and Meta sound. These companies are selling this as a kind of advertisement. Look, how good our models are already. However, this is the completely wrong discussion.

Let's imagine this wasn't about AI companies, but about companies researching biological weapons. If a potentially humanity-threatening weapon were to simply escape from a secure laboratory, the shock would be great. And rightly so. Such companies could ultimately even be shut down. But with AI companies, it doesn't seem to have that effect so far.

OpenAI and Anthropic are preparing for their initial public offerings. The CEOs Sam Altman and Dario Amodei never tire of emphasizing the power of the systems they have developed. As powerful as the invention of the internet? Certainly. As influential as the first steam engines? Probably. So powerful that future AI systems could wipe out humanity? Definitely. At least, if you believe the CEOs.