Podcasts

The DHS inspector general will investigate forced CISA reassignments, Canada hacked a ransomware gang, Taiwan charges two executives with helping Chinese hackers, and new vulnerabilities can disable Hoymiles solar panels.

In this wholly sponsored Soap Box edition of the podcast Patrick Gray chats with Damien Lewke, the CEO and founder of Nebulock, about the future of threat hunting and detection.

Damien spent a decade in the EDR and MDR space before founding Nebulock in 2024. It started off as an AI-powered threat hunt platform but has evolved into a broader security data platform that can answer questions, drive hunts and drive detections.

This product is engineered around the idea that a lot of security is a data problem. So, if we accept this premise, how do we solve security? And how much of that solution is about agents, vs building a good graph? And if you’re going to build a good graph, do you want to build it for a person to use, or an agent to use?

This is truly a conversation for the security nerd’s nerd. Enjoy!

This episode is also available on YouTube

In this edition of Between Two Nerds Tom Uren and The Grugq talk about why we haven’t seen an explosion of devastating hacks even though AI has been used to discover lots and lots of bugs.

This episode is also available on YouTube.

A European MP’s phone was infected by Pegasus spyware, Android drops its PIN guessing limit from 1,800 attempts to 20, Alibaba bans employees from using Claude at work, and there’s a new vulnerability in the Linux kernel.

FatFs bugs enable physical access attacks on industrial equipment, a clever password spraying attack bypasses M365 MFA, an AI agent is deploying ransomware in live attacks, and a webinar platform sues two security firms over bad IOCs.

Tom Uren and James Wilson talk about Chinese AI labs stealing the special sauce of American AI models in ‘distillation attacks’. These attacks are fed by a grey market in which Chinese consumers buy access to American models, where one of the byproducts is logs of user requests and responses. These make wonderful inputs into distillation attacks and the whole market might be subsidised by Chinese AI Labs paying for these logs.

They also discuss the possibility that last year’s hack of Jaguar Land Rover was caused by a group of Russian hackers. Was it Russians? Was it state-directed or endorsed? Who knows, but even the possibility that it was has some benefits for the Russian state.

This episode is also available on YouTube

An anonymous researcher has dropped a giant cache of zero-day exploits, a sensitive DHS network got hacked, the US Supreme Court restricts geofence warrants, and security firm Huntress has denied accusations of a malicious insider.

On this week’s show Patrick Gray, Adam Boileau and James Wilson discuss the week’s cybersecurity news. They cover:

• Anthropic’s Fable 5 returning while OpenAI’s GPT-5.6 gets thrown in model jail
• Distillation, cheap tokens, and AI chat harvesting is an industry in China
• Edge becomes a lolbin via a new malicious extension
• An Iranian APT boss’s vacation in a beautiful place goes wrong
• Much, much more!

In this week’s sponsor interview Daf Stuttard and Katie Warren from Portswigger pop along to talk about how they built an AI security testing product that people would actually feel comfortable using.

This episode is also available on YouTube.

In this podcast episode James Wilson chats with Karsten Nohl about his research into using local LLMs to replace cloud AI in security code reviews.

In essence, Karsten created a hybrid code reviewing system where both cloud and local models are used to orchestrate, triage outputs, and write reports. In this system, only the local LLMs have source code access, with the cloud models used to manage the local models.

In this “source-local” review technique, the source code never leaves the local endpoint, which is a requirement for some reviews. But funnily enough, Karsten was able to use this system to generate findings that were as impressive as when using frontier models directly.

In a nutshell, Karsten proved it’s possible to use locally-hosted, open-weight models running on commodity hardware to produce findings comparable to those discovered by frontier cloud models.

This episode is also available on YouTube.

In this edition of Between Two Nerds, Tom Uren and The Grugq discuss whether cyber organisations should actually be separated from Signals Intelligence organisations. The Grugq argues that having cyber expertise subordinate to intelligence collection means that many opportunities are never explored.

This episode is also available on YouTube.