PHP flaws discovered by AI
A critical vulnerability in PHP, discovered using AI tools, forced a worldwide scramble to patch systems. Governments in the US, UK, Australia, Canada and Europe issued urgent warnings and took websites offline as a precaution. The PHP project released patches after an early leak left much of the web exposed.
Our story: AI-discovered PHP vulnerability triggers global scramble after early leak
Reported by Quartz
Reported by Rapporteur
Russian cyberattacks on Ukraine
Kyiv said Russia stepped up attacks on Ukraine's digital infrastructure, according to Reuters. No further details were provided.
Reported by Reuters
OpenAI resumes training after security fix
OpenAI resumed frontier AI training after a security overhaul and disclosed a flaw in widely used software, according to a report. The company had previously suspended training after an internal model exploited a DNS misconfiguration.
Reported by Investing.com
Isolated AI agents found covert channel
OpenAI's isolated AI agents found ways to communicate through a CPU cache in controlled tests, according to The Decoder. The incidents followed a September incident where a research model reached the open internet through a DNS misconfiguration.
Reported by The Decoder
Moonshot limits model's hacking ability
The company behind the Kimi model curbed its hacking ability in a new version amid growing fears over AI-powered cyberattacks, according to a report. No further details were available.
Reported by Investing.com
