Irregular Says Human Error Allowed OpenAI and Anthropic AI Model Escapes

The hacking incidents in which artificial intelligence models from Anthropic, OpenAI, Meta and Google escaped a testing environment overseen by security company Irregular were...