OpenAI fashions going rogue, the primary documented agentic ransomware operation, and an emergent AI-driven provide chain risk made for a packed July roundup
31 Jul 2026
With July coming to a detailed, ESET Chief Safety Evangelist Tony Anscombe appears to be like again at a few of the high cybersecurity tales that made the information over the previous month and gives insights that they might maintain to your personal cyber-defenses. This is a few of what caught Tony’s consideration this month:
- OpenAI fashions went rogue and autonomously broke into AI collaboration platform Hugging Face in what OpenAI described as “an unprecedented cyber incident”.
- Researchers at Sysdig have documented what they assess to be the primary case of an end-to-end ransomware operation executed by an agentic risk actor. The corporate named it JADEPUFFER.
- Cybercriminals are benefiting from a brand new massive language mannequin (LLM)-driven assault vector referred to as “phantom squatting” by buying domains linked to official manufacturers, then utilizing them to intercept visitors directed there by AI techniques.
What can organizations do to cease phantom squatting from harming their manufacturers, and what different classes do these incidents maintain for defenders? Watch Tony’s video to seek out out and remember to try the June 2026 version of his month-to-month safety information roundup for extra insights.
Earlier than you go, study concerning the first AI-powered ransomware, named PromptLock and found by ESET researchers final yr. To study extra about cutting-edge AI protection layers, learn the AI at ESET white paper.
