First findings from Project Glasswing

First findings from Project Glasswing

While Anthropic has restricted Mythos access to its Project Glasswing partners, it has always maintained that lessons from Glasswing would be shared with the broader cybersecurity community. Now, those lessons are starting to roll out. This week, on Security Intelligence, panelists Dustin “EvilMog” Heywood, Kimmie Farrington and Curtis Pitts discuss Cloudflare’s recent write-up on its adventures with Mythos so far. We discuss what separates Mythos from other AI vulnerability hunters, Cloudflare’s agentic harness and whether “speed” is the wrong way to think about AI cybersecurity tools. Then: A CISA contractor accidentally exposed a repo full of cloud keys, passwords, tokens and other credentials to the public web on GitHub. It’s a case study in identity and access management mistakes and supply chain vulnerabilities—and there’s a lot to learn from ti. Finally, we look back on L0pht Day, 1998, when a group of Boston-area hackers warned Congress about the fundamentally inadequate security measures of the early internet. Have we made any progress since then? Maybe not as much as you think. All that and more, on Security Intelligence.

OpenAI’s Daybreak and Mistral’s Mythos competitor

Between OpenAI Daybreak, Microsoft MDASH and Mistral’s Mythos competitor, it’s been a big week for AI-powered vulnerability management. But are these tools all they’re cracked up to be? This week on Security Intelligence, Nick Bradley, Diego Matos Martins and Nikki Robinson discuss three bold moves in the AI vulnerability scanner space: OpenAI unveiled Daybreak, its frontier AI for cyber defense program, Microsoft revealed its multi-agent vulnerability hunting system, MDASH, and French AI startup Mistral is reportedly building its own cybersecurity-focused model to fill the gap left by the lack of access to Anthropic’s Mythos in Europe. Speaking of Mythos: curl developer Daniel Stenberg got to try it himself (sort of), and his verdict was measured, to put it kindly. But despite this—and the fact that AI slop reports drove curl to shut down bug bounties earlier this year—Stenberg is far from anti-AI. We dig into why. Finally: TeamPCP released the source code for Shai-Hulud, the notorious worm behind a surge of npm supply chain attacks. They're even running a dark web contest to crowdsource new attack variants. What’s it all mean for defenders? All that and more on Security Intelligence. Segments: 00:00 -- Intro 1:17 -- Daybreak, MDASH and Mistral 11:31 -- Curl dev tries Mythos 20:57 -- Shai-Hulud goes open source 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. Follow the Security Intelligence podcast on your preferred platform: https://www.ibm.com/think/podcasts/security-intelligence [https://www.ibm.com/think/podcasts/security-intelligence]

LLMjacking: How hackers steal your AI API keys and stick you with the bill

AI tools can turn a team of three developers into a fully functioning company. They can also push that company to the brink of bankruptcy. On this week’s Security Intelligence, we talk LLMjacking: Hackers steal your AI API keys and then rack up massive bills, even blowing past usage caps in some cases. One small startup saw its typical bill balloon from $180 a month to $82,000 in two days. We chat about what makes AI API keys vulnerable and how we can tighten our defenses to keep these vital credentials safe. Then we get into how AI is transforming adversary simulation and red teaming, and why the human is still the most important part of the loop. Finally, CISA is considering cutting the federal patch window from two weeks to three days. Can we actually move that fast? Segments: 00:00 – Intro 1:15 -- What is LLMjacking? 12:29 -- AI and adversary simulations 22:09 -- Can we patch faster? 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. Follow the Security Intelligence podcast on your preferred platform: https://www.ibm.com/think/podcasts/security-intelligence [https://www.ibm.com/think/podcasts/security-intelligence]

Claude Security’s public beta, OpenAI’s five-point plan and cybersecurity’s Y2K moment

Between Mythos, GPT-5.4-Cyber, Claude Security’s public beta and OpenAI’s new five-point plan for cyber defense, it seems like cybersecurity is top of mind for the major AI players today. Why—and why now? On this week’s episode of IBM Security Intelligence, Dustin “EvilMog” Heywood, Omari Jones and Kimmie Farrington discuss what CrowdStrike has called “cybersecurity’s Y2K moment.” As the major AI players roll out security-focused solutions—and sophisticated AI tools are weaponized by threat actors—we need all-hands on deck to avert disaster. But will we? Plus: The Coalition for Secure AI’s framework for AI identities and Copy Fail, a newly discovered Linux flaw with a potentially massive blast radius. All that and more on Security Intelligence. Segments: 00:00 -- Intro 1:11 - Cybersecurity’s Y2K moment 10:52 -- Framework for AI identity 22:23 -- Copy Fail 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. Follow the Security Intelligence podcast on your preferred platform: https://www.ibm.com/think/podcasts/security-intelligence [https://www.ibm.com/think/podcasts/security-intelligence]

Is open source safe? Featuring Mixture of Experts

Is open source good? Bad? Some secret third thing? Is this a silly question to even ask? In this special crossover episode of Security Intelligence and Mixture of Experts, we bring together AI and security experts to address one of the thorniest questions in tech right now: How do you enjoy the unique benefits of open source AI while managing its very real risks? MoE stalwarts Gabe Goodhart and Martin Keen join SI all-star Jeff Crume to dig into: * Why open source is foundational to AI innovation * Security concerns of both proprietary and open source AI infrastructure * The difference between "secure" and “securable” * And a whole lot more! Along the way, we hash out a robust, nuanced picture of the relationships between AI, security and open source. Go beyond the buzzwords to what really matters on this week’s episode of Security Intelligence. 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. Follow the Security Intelligence podcast on your preferred platform: https://www.ibm.com/think/podcasts/security-intelligence [https://www.ibm.com/think/podcasts/security-intelligence]