Breaches, vulnerabilities and the people cleaning up after them. Reporting for defenders, not for panic.
A threat actor tracked as Storm-3168 is leveraging autonomous AI agents to orchestrate attacks against Azure tenants, automating reconnaissance, credential harvesting, and destruction of cloud infrastructure across multiple resource types.
Following recent sandbox escapes at major AI labs, Nvidia has unveiled a system pairing its OpenShell runtime with a BlueField-4 based watchdog to contain rogue AI agents through deterministic policy enforcement and network-level controls.
The cryptocurrency exchange has begun restoring withdrawal services after attackers compromised backend wallet infrastructure and drained over $350 million across multiple blockchain networks.
U.S. federal agencies must secure systems against two critical Citrix NetScaler vulnerabilities by September 30 after active exploitation was confirmed in the wild.
Two remote code execution vulnerabilities in Citrix NetScaler are being actively exploited in the wild, prompting cybersecurity agencies and IT providers to privately warn organizations to take appliances offline before patches arrive next week.
Two compromised third-party GitHub Actions were restored to active status but continued serving malicious payloads for over a week, putting downstream workflows at risk of executing infected code.
OpenAI has identified 53 instances where its AI agents inadvertently transmitted user-provided images to external image-hosting platforms, prompting the company to strengthen safeguards around agent behavior and data handling.
The secure file-sharing vendor instructed customers to take systems offline for six hours after receiving law enforcement intelligence about a potential weekend attack.
