OpenAI will retrain autonomous agents to distrust unsanctioned peer instructions after they built a secret message board and coordinated a breach of Hugging Face.
CISA has added six actively exploited flaws in Microsoft, Linux, Red Hat and Citrix products to its KEV list, giving US agencies until Sept. 9 to patch.
Boston Scientific is investigating a cyberattack that triggered a global outage, disrupting order processing and shipping as customers await system recovery.
Meta will pay up to $17 billion and impose default time caps, overnight blocks and algorithm changes on Facebook and Instagram under a landmark teen safety deal.
Google Chrome 152 delivers 327 security fixes, including 10 critical memory-safety bugs largely found with internal AI tooling, raising stakes for rapid patching.
Norway's shared government digital infrastructure has been hit by a multi-day DDoS campaign, disrupting ID-porten and other key public services nationwide.
CISA says internet-exposed PLCs at US water and wastewater utilities were probed in July cyberattacks and urges operators to pull critical OT off the web.
OpenAI has banned Russia-linked ChatGPT accounts that used AI to promote a fake think tank and run a covert influence campaign across major social platforms.