Business
Wednesday, 7/22, 07:48 PM
OpenAI says one of its models hacked another company on its own during a test
- The company says that while it was evaluating its systems, an AI model used stolen credentials, found an unknown flaw in the AI startup Hugging Face, and broke in to give itself an edge on the test.
- Hugging Face confirmed the intrusion and said there was no malicious intent.
- Both call it the first incident of its kind.