The Department of Know: GemStuffer attack, AI SBOMs, and AI-created zero-days

Cybersecurity Headlines34mMay 15, 2026

Get the full intelligence

Search transcripts, export clips, track mentions, and explore all topics from “The Department of Know: GemStuffer attack, AI SBOMs, and AI-created zero-days” inside PodZeus.

AI-Generated Summary

This episode of Cybersecurity Headlines dives into a series of high-impact cyber threats and policy developments, starting with the 'GemStuffer' attack on RubyGems, where malicious actors used the package manager as a dead drop for exfiltrated data from UK government websites—marking a novel use of legitimate infrastructure. The discussion then shifts to AI-driven zero-day exploits, with Google uncovering a cybercrime group using LLMs to develop a sophisticated bypass of two-factor authentication in an open-source admin tool, underscoring the accelerating threat of AI-generated attacks. The panel also examines broader geopolitical and policy shifts, including the EU’s potential ban on U.S. cloud providers for sensitive government data, the launch of AI Software Bill of Materials (SBOMs), and the UK’s plan to modernize its outdated Computer Misuse Act. Both CISOs emphasize that while the AI apocalypse is real, the key to survival lies not in fear but in adaptation—rethinking security programs, embracing faster response cycles, and recognizing that traditional models are no longer sufficient in an era of AI-powered threats. The episode closes with a call to action: security teams must evolve from 'faster horses' to 'cars'—reimagining their entire approach to resilience and innovation. Key takeaways include: (1) Threat actors are increasingly abusing legitimate platforms like RubyGems and HuggingFace as covert infrastructure; (2) AI-generated zero-days are no longer theoretical—they’re already in use, and detection must evolve; (3) Policy shifts like AI SBOMs and data sovereignty laws will force architectural changes across global tech ecosystems; (4) Security teams must prioritize adaptability over speed, rethinking their programs entirely; and (5) The cost of AI tools is not the barrier for attackers—defenders must prepare for a new era of rapid, automated threats. The tone is urgent but hopeful, emphasizing that while the landscape is changing, the industry has the tools and momentum to adapt.

Key Takeaways
1

Threat actors are repurposing legitimate platforms like RubyGems and HuggingFace as covert data exfiltration channels, bypassing traditional detection.

2

AI-generated zero-day exploits are now active, with attackers using LLMs to develop sophisticated, well-documented vulnerabilities—marking a new phase in cyber warfare.

3

Global policy shifts like EU data sovereignty rules and AI SBOMs will force major architectural and compliance changes across organizations.

4

Security programs must evolve from incremental improvements to fundamental rethinking—'you don’t need a faster horse, you need a car.'

5

The real threat isn’t just the technology—it’s the assumption that reputation, ratings, or trust signals are reliable indicators of safety.

Chapters
0:00
10 min

Opening: The AI Apocalypse & Security Realities

Host Rich kicks off the episode with a lively intro, setting the tone for the week’s cybersecurity news. He welcomes guests Gary Chan (CISO, SSM Health) and Peter Liebert (CISO, SalesLoft), emphasizing the theme of adaptation in the face of emerging threats. The discussion begins with a focus on the looming AI-driven cyber threat landscape, with Peter joking about preparing for the 'AI apocalypse' and Gary highlighting the importance of frontline IT staff knowledge.

10:00
10 min

GemStuffer Attack: Abuse of RubyGems as a Dead Drop

It's not about the capability of the tool—it's about how it's being used. That's the new threat model.

Highlight
20:00
10 min

AI-Generated Zero-Days: Google Uncovers LLM-Powered Exploits

The code looked too good. That should’ve been a red flag—no human engineer writes documentation like that.

Highlight
30:00
10 min

Policy Shifts: EU Data Sovereignty, AI SBOMs, and UK Law Reform

This isn’t just about data—it’s about culture, influence, and the future of global cohesion.

Highlight
40:00
17 min

Closing: Rethinking Security in the Age of AI

The episode concludes with actionable advice from both guests. Peter urges rapid adaptation, emphasizing that existential threats are finally breaking through organizational inertia. Gary delivers a powerful metaphor: 'You don’t need a faster horse—you need a car.' The hosts wrap up with a reminder of the show’s live Friday stream and an upcoming event on AI-powered pen testing.

High-Impact Quotes
You don’t need a faster horse—you need a car.
Gary Chan54:10
Viral: 95.0
The code looked too good. That should’ve been a red flag—no human engineer writes documentation like that.
Peter Liebert45:15
Viral: 90.0
This isn’t just about data—it’s about culture, influence, and the future of global cohesion.
Gary Chan35:30
Viral: 88.0
Speakers

Host

Rich

Guests

Gary ChanPeter Liebert
Topics Discussed
AI-Generated Cyber Threats95%Security Program Evolution92%Supply Chain Security90%Zero-Day Exploits88%Data Sovereignty and Policy85%Software Bill of Materials for AI80%Social Engineering and Trust Signals78%Threat Actor Ecosystems75%
People & Brands

Peter Liebert

person

18xPositive

Gary Chan

person

15xPositive

RubyGems

product

12xNegative

GemStuffer

other

8xNegative

HuggingFace

other

6xNegative

Google Threat Intelligence Group

organization

5xPositive

CISO Series

media

5xPositive

European Union

organization

5xNeutral

Department of Homeland Security

organization

4xNeutral

OpenAI

organization

4xNeutral

Get the full intelligence

Search transcripts, export clips, track mentions, and explore all topics from “The Department of Know: GemStuffer attack, AI SBOMs, and AI-created zero-days” inside PodZeus.

Start discovering podcast insights today

Start with a 7-day trial and explore a growing catalog of popular podcasts. No credit card required.

No credit card required • 7-day trial • Cancel anytime