security teams

AI agents are increasingly filing automated security vulnerability reports against open source projects, overwhelming maintainers and straining established disclosure practices. The reports, often generated by autonomous tools hunting for bugs, can be noisy or low quality, forcing projects to rethink triage, disclosure norms and coordination. Security teams and maintainers are debating how to adapt responsible disclosure processes to a flood of machine-generated findings.

The Akira ransomware group has published new posts on its leak site, apparently claiming attacks on two organisations, including a technology-sector company listed as 'DPL Group'. The disclosures were flagged by cyber threat intelligence monitors who track ransom group blogs for new victim claims. Akira remains an active ransomware operation, and new leak-site entries are closely watched by security teams assessing exposure of their own suppliers and partners.

Commentators are arguing that AI's real impact on cybersecurity is not novel attacks but the collapsing cost of running existing ones. Attacks that were previously too expensive to launch against smaller companies are now affordable, meaning businesses of any size can become viable targets. The argument is that nothing technically new was required for this shift, only automation lowering the price of techniques security teams already know about.

A vulnerability in the Linux kernel's Reliable Datagram Sockets networking subsystem has been resolved, with the fix addressing TCP stream corruption that could occur when large memory pages are in use. The advisory, published 25 September 2026 and updated 3 October, affects the Linux kernel and a range of related products, drawing attention from security teams tracking kernel updates.