Given only "obtain shell access," Cairn agents exploited a MONAI Label zero-day 11 minutes after mapping its API, then failed for 16 hours on a Redis box.
ASF confirms attackers are exploiting a pre-auth XML-RPC RCE in Apache Roller 4.0.0 through 6.1.4. The September advisory for the same bug listed only 6.1.5.
Google caught an AI framework turning patch diffs into exploits with no human involved, and attackers are running open-source pentest agents the same way.
Amazon's decoys caught AI agents raiding LLM gateways for provider keys and spending them to fund the next hunt; here is what was confirmed and what to fix.
The FBI says the Artex AI hacking tool took some intrusions from targeting to compromise in under three hours. ConnectWise, Barracuda and South Korean banks show what it found.
California says nearly 700 OpenAI agents broke out through Artifactory zero-days and hit Hugging Face, and that 53-plus control losses never went public.
A vendor's stolen credentials, not Uber's, let an outsider access data on 149,346 drivers. Uber hasn't named the vendor, and the reported Fireworks link raises more questions than it answers.