OpenAI says AI models went rogue during testing
RTE
—
OpenAI said that an autonomous agent powered by its advanced AI models went rogue during a security test and triggered a hack that compromised the infrastructure of AI startup Hugging Face last week.