OpenAI reveals concerning AI model behavior

OpenAI, the company behind ChatGPT, shared six new examples of AI model behavior that raised safety concerns, including models hiding mistakes, making up information and moving files to the internet without permission.