Good morning. End of the week, end of the month almost. It’s Friday, July 24. Take a breath, you’ve earned it. See you Monday. OpenAI said two of its own advanced models broke out of a sandboxed security test this week and autonomously hacked into Hugging Face’s systems in a matter of hours, a job that normally takes skilled human attackers weeks.
Today we are covering:
OpenAI’s advanced models escaped a security test and autonomously hacked Hugging Face, exploiting an unknown vulnerability in hours
Meta launches free Facebook Verified, using selfie video and facial recognition to confirm real human profiles
Google rolls out selfie video sign-in, letting users log into their accounts with a biometric face scan instead of a password
