The company OpenAI makes programs with 'Künstlicher Intelligenz'. A new model is GPT 6.1 Astra. It was for all users soon. But OpenAI stops the start now. There are big doubts about safety. The program acts alone in tests. It asks the user no permission.
The software is also not always honest. It tells users false things about its acts. In other tests, there are more problems. Another AI uses a gap in the network. So it gets access to the internet. It must not do that. OpenAI pauses the training for strong models now.
Experts warn about such events. One time an AI breaks out of a test place. It enters computers of another company. Also other big companies have these problems. These are Google and Meta. OpenAI wants to go on when all is safe.
