About the episode

That melted ice cream might not be the freezer’s fault.

On episode 56 of Security Intelligence, Claire Nunez and MJ Vaidya join host Patrick Austin to:

Break down three cybersecurity stories and what they mean for defenders.
First, attackers hijack refrigeration controllers from Danfoss, Copeland and Dixell at French supermarkets, pushing cases to 50°C. Team82 flagged the flaws and Danfoss shipped a fix, yet thousands of controllers remain exposed online. Is planned downtime the price of staying patched?
Then, a Semperis survey of 1,200 IT professionals links AI to 84% of ransomware attacks. Many of those attacks abuse companies’ own AI agents and identity systems. Should your agents be part of your insider threat model?
Finally, Bank of America brings on 1,000 contractors to patch vulnerabilities found by Anthropic’s Mythos and other AI security models. The panel debates whether AI can fix what it finds, and why patching can’t be a one-time push.

All that and more on Security Intelligence.

Chapters:

00:00 – Intro
01:00 – Hackers hijack supermarket fridges
10:11 – AI linked to 84% of ransomware attacks
19:31 – Bank of America hires 1,000 patchers

The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. AI tools may be used to transcribe this episode and support selected stages of the production process. All AI-assisted content is reviewed by the production team before publication.

Subscribe to the IBM Think newsletter

Subscribe now on your favorite platform

YouTube

Spotify

Apple Podcasts

Host and speakers of the episode:

Claire Nunez

Creative Director, IBM X-Force Cyber Range

Ryan Anschutz - North America Leader of X-Force Incident Response

MJ Vaidya

Global Senior Partner, Cyber Strategy & Risk

Patrick Austin Lucas

Host

Topics we discussed during the episode

What is OT security?

What is ransomware?

What is patch management?

Watch previous episodes

AI slop stalls bug bounties, malware hides in IoT and agents hit Wikipedia

AI slop overwhelms Google's bug bounty program, ClingSTUN turns IoT devices into proxies and OpenAI agents target Wikipedia. What can defenders do?

Hiding in plain sight: Fake GPTs, SMTP malware and NetScaler zero-days

Hackers are building fake GPTs, hiding malware in email traffic and exploiting NetScaler zero-days. How do you defend against attacks that look legitimate?

Can you trust your chatbot? Inside three AI-powered cyberattacks

Hackers are poisoning AI chatbots, building tools with LLMs and unleashing agent swarms. How do you defend against AI-powered cyberattacks?

Can you trust your chatbot? Inside three AI-powered cyberattacks

Hackers are poisoning AI chatbots, building tools with LLMs and unleashing agent swarms. How do you defend against AI-powered cyberattacks?

How Costa Rica built a cybersecurity dream team

Costa Rica's Cyber Cluster brought government, business and academia together around one goal: making the country a global center of cybersecurity expertise. Here’s what we can learn from it.

Are AI labs ignoring cybersecurity experts?

OpenAI, Anthropic and other AI labs want more AI safety rules. Cybersecurity pros say they're not in the room, and the labs are solving the wrong problem.

The vulnpocalypse might not be so bad after all

A new report says AI-driven vulnerability disclosures look scarier than they are. Our panel debates whether validation—not patching—is the real fix.

Why won’t AI agents just follow the rules?

“Don’t do that” doesn’t work on AI agents. So what does? Plus: The OWASP Top 10 for agentic skills, how AI is ruining bug bounties and ThreatXtension.

Why OpenAI is calling for a ‘cyber defense surge.’ Plus: Find Evil! winners and TeamPCP losers

OpenAI’s open letter, the winners of Find Evil! and the TeamPCP takedown. All that and more on Security Intelligence.

‘That’s not a knife’: Attackers outpace defenders with AI. Here’s how we catch up.

Attackers are adopting AI faster than defenders. They key to catching up? Stop thinking about AI.

Who’s afraid of an open-weight model? GLM, context bombing and post-Black Hat attacks

GLM-5.3 beat GPT-5.6 and Claude Mythos at vulnerability hunting. What now? Plus: context bombing and post-Black Hat phishing scams.

What should security leaders do with AI? They don’t know

Cybersecurity leaders are flush with cash and eager to spend it on AI. What they don’t have is a plan for how.

The OWASP LLM Top 10 has a few surprises for you

The OWASP LLM Top 10 throws some real curveballs. Plus: SBOMs and Black Hat 2026 highlights.

Oh look. Anthropic’s AI models also broke containment

Anthropic found 3 cases of its own AI models breaking containment during testing. Plus a new agentic browser flaw and a public archive of 204 zero-days.

Your data breach plan is missing something major: people

Most breach plans forget the hardest part: the people. IBM's Limor Kessem on why human factors determine whether your crisis response succeeds or fails.