This Week in AI Security - 30th April 2026

In this episode for April 30, 2026, Jeremy breaks down a week where the "human-in-the-loop" failed spectacularly. From a production environment deleted in just nine seconds to "Abliterated" models providing kidnapping instructions to Congress, the risks of autonomous AI agents are no longer theoretical. They are live.

Key Episode Highlights:

Abliterated Models on Capitol Hill: OpenAI and Anthropic briefed House lawmakers on "abliterated" models - versions with safety guardrails stripped - demonstrating how they can provide step-by-step instructions for criminal acts.
Entra ID Hijacking: Researchers at Silverfort discovered that the new "Agent ID" role in Microsoft Entra ID can be exploited to hijack service principals, leading to a full Global Admin takeover.
The 9-Second Disaster: An AI agent at PocketOS, attempting to fix a staging environment, fetched production credentials and deleted both the production environment and its backups in under ten seconds.
LiteLLM SQL Injection: A critical vulnerability in the LiteLLM gateway saw targeted exploitation within 36 hours of disclosure, specifically aiming for provider API keys.
Vercel Breach Update: The recent Vercel data breach is traced back to a "Luma Stealer" malware infection at a third-party AI analytics partner.

Episode Links

Is your organization part of the 82% with unknown AI agents running on your network? Don't wait for a "9-second deletion" event. Get full visibility into your AI agents today.

Book your FireTail demo: https://www.firetail.ai/schedule-your-demo

Denne episoden er hentet fra en åpen RSS-feed og er ikke publisert av Podme. Den kan derfor inneholde annonser.

Episoder(127)

David Kerber of Act Security

In this episode of Modern Cyber, Jeremy is joined by David Kerber from Act Security and Cloud Copilot to explore the complex, heavily misunderstood world of AWS IAM. David dismantles common misconcept...

18 Aug 37min

This Week in AI Security - 13th August 2026

Fresh off Black Hat and DEF CON, Jeremy raises the bar on which stories make the cut and walks through the most compelling disclosures from a packed couple of weeks. The dominant theme: agents pursuin...

13 Aug 15min

This Week in AI Security - 6th August 2026

Recorded from the sidelines of hacker summer camp, Jeremy runs through a packed week spanning Black Hat, B-Sides, and DEF CON. The theme keeps repeating: prompt injection is always possible, and it is...

6 Aug 21min

This Week in AI Security - 30th July 2026

The final episode before Black Hat, and Jeremy keeps it tight with a few quick hits before settling into the week's biggest theme: identity, visibility, and the open-versus-closed model debate. This w...

30 Jul 17min

This Week in AI Security - 23rd July 2026

A lighter week on volume that Jeremy uses to go deep on two of the most significant stories of the year so far. The episode opens with quick hits on export-control pressure spreading to OpenAI's model...

23 Jul 23min

This Week in AI Security - 16th July 2026

Another lighter week that lets Jeremy slow down and dig into the stories that matter most. The theme running through this episode: the tooling and plumbing around AI keep proving to be the real attack...

16 Jul 15min

This Week in AI Security - 9th July 2026

A quieter summer week on the news front, which gives Jeremy room to dig deeper into a handful of stories that all circle the same theme: the tooling and infrastructure around AI keep proving to be the...

16 Jul 12min

This Week in AI Security - 2nd July 2026

A lighter week on volume, which gives Jeremy room to go deeper on a set of stories that all reinforce trends we've been tracking for months. The through-line: prompts keep showing up in places nobody ...

2 Jul 12min