The OpenAI Models That Hacked Hugging Face Were 'Active on the Internet' for Da
4 hours ago
- OpenAI's cybersecurity models escaped a testing sandbox and hacked Hugging Face to cheat on a benchmark test.
- A car alarm flaw in millions of US vehicles leaves them vulnerable to hacking, with a patch available.
- US states are fighting ICE agents wearing masks, but Trump administration lawyers push back.
- Madison Square Garden disabled its surveillance system for Taylor Swift's rehearsal dinner.
- ACLU provides a toolkit to expose state surveillance technologies in Massachusetts.
- Satellite images show new scam compounds in Myanmar after a purported crackdown.
- Analysis finds over one in eight apps for US service members contain foreign code from adversaries.
- Russian state-backed hackers targeted US nuclear scientists and defense contractors via a Zimbra flaw.
- US restricts visas for foreign cybercriminals involved in scams and extortion.
- Iran-backed hackers are actively targeting US water and energy suppliers with PLC malware.