U.S. courts will soon publicly report how often judges authorize spyware and hacking tools to intercept communications, giving Americans a clearer view of government surveillance.

The FBI is warning that criminals are hijacking online accounts to steal intimate photos and videos, using familiar tactics including phishing, fake support messages, and password attacks.

Bumble now lets anyone in a match send the first message and gives the recipient 72 hours to respond, replacing its previous 24-hour window.

OpenAI’s GPT-5.6-Cyber handles advanced security requests its standard models often refuse, arriving as recent evaluations show autonomous AI agents crossing intended boundaries during real cybersecurity testing.

OpenAI is pausing some work on Astra after testing found the AI could identify and exploit software vulnerabilities without human intervention.

UK children reported 420 cases of explicit deepfakes in six months, surpassing the previous year's total as AI makes abusive imagery easier to create.