The Adversarial Podcast

The Adversarial Podcast

di Jerry Perullo, Sounil Yu, Mario Duarte
LavoroTecnologia
Stagione 4
S4E24 – Dream’s agentic attack report, cyber privateers, Taiwan’s internet drill
00:00 Cold Open: The Offensive-Cyber Incentive Problem 00:31 Welcome and the macOS Screen Sharing Flaw 03:14 Patching the Humans After DEF CON 12:14 CMDBs, Shadow IT, and Just-in-Time Context 19:59 Cloudflare’s Markdown for Agents 23:47 Taiwan’s Live Internet-Throttling Drill 33:46 Can an AI Agent’s Own Logs Be Forensic Evidence? 38:57 Cyber Privateers and the New Offensive-Cyber Program 43:32 How Commercial Hack-Back Might Work 51:48 The Rogue Delta Wi-Fi Network 57:33 AI Red Teaming and Automated Remediation 59:09 Replacing Vendor Questionnaires with Real Testing 1:02:05 When Red Teaming Creates Defensive Bloat 1:03:46 Pen Tests Find Flaws; Red Teams Pursue Outcomes 1:06:25 Closing Thoughts Stories and resources Inside a Multi-Agent AI Framework Used to Compromise Government Entities in Asia — Dream Research Labs OpenAI and Hugging Face partner to address security incident during model evaluation — OpenAI Taiwan briefly slows its mobile internet as part of defense drill — AP Private companies authorized to conduct offensive cyber operations — TechRadar Delta flight Wi-Fi tampered with after DEF CON — ITPro United flight turns around over a suspicious Bluetooth device name — NPR/CapRadio Markdown for Agents — Cloudflare Critical macOS Screen Sharing flaw — Tom’s Guide Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (CISO, https://www.whirlai.com/) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E24
18 ago 2026
1:06:53
S4E23 – AI Agents Escape Multiple Frontier Labs
Chapters 00:00 Introduction to AI security challenges 02:05 Recent hacking incidents involving Hugging Face and Anthropic 04:01 How AI models find ways to cheat and bypass constraints 05:56 The challenge of containment and governance in AI safety 08:00 Lessons from recent AI security breaches 10:01 The role of human oversight in AI security testing 12:03 Cost and effectiveness of offensive AI security measures 13:54 Implications for critical infrastructure and national security 16:03 Policy and regulatory impacts on AI safety 17:52 Future strategies for AI containment and defense 20:11 Conclusion and key takeaways HuggingFace: Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident Hugging Face reconstructs an autonomous intrusion involving approximately 17,600 actions over a multiday campaign. Anthropic: Investigating three real-world incidents in our cybersecurity evaluations After reviewing 141,006 cybersecurity-evaluation runs, Anthropic identified three incidents in which Claude reached real organizations through evaluation infrastructure that had been mistakenly connected to the internet. The incidents spanned six runs and three models. Anthropic reached two of the affected organizations, neither of which had detected the activity before being notified. Anthropic did not disclose token usage or inference costs for these intrusions. Anthropic: Discovering cryptographic weaknesses with Claude Anthropic reports that Claude Mythos Preview progressed from finding implementation flaws in cryptographic libraries to identifying mathematical weaknesses in cryptographic algorithms themselves. Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (CISO, https://www.whirlai.com/) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E23
4 ago 2026
1:06:00
S4E22 – HuggingFace compromised by agentic attack, Gold Eagle program
00:00 The Adversarial Podcast 00:58 Hugging Face’s AI-driven incident disclosure 03:48 What makes an attack “AI-enabled” 06:04 Exploits, vulnerabilities, and bespoke code execution paths 10:03 AI attackers vs. AI defenders 11:19 Verification asymmetry: attackers vs. defenders 13:03 Detective controls, red teaming, and breach simulation 16:41 Why AI defenders matter 26:25 Gold Eagle / national coordination of vulnerability discovery 28:25 The real bottleneck is remediation, not discovery 37:04 CMMC and the cost of certification 42:15 Is certification effective, or just paperwork? 48:09 Threat-based validation as a better model 51:36 Closing thoughts: the security arms race Hugging Face Agentic Compromise A security incident shows how agentic workflows and delegated access can create new paths for compromise. Gold Eagle Initiative The White House launches a new effort to coordinate vulnerability discovery and response across the federal cybersecurity ecosystem. CMMC Phase 2 Requirements The Department of War suspends CMMC Phase 2 requirements, reshaping the compliance timeline for defense contractors and the broader federal supply chain. Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, stealth startup) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E22
21 lug 2026
52:40
S4E21 - Travel Security, AI Defense Matrix, Startup Security
In this episode, Jerry, Mario, and Sounil delve into cybersecurity challenges related to travel, threat models, AI security, and best practices for startups. They explore practical strategies for managing security risks in a rapidly evolving digital landscape, emphasizing the importance of threat modeling, secure coding, and organizational priorities. 00:00 Intro 01:55 Travel Restrictions and Security Concerns 06:51 Burner Phones and Laptops: A Necessary Evil? 09:50 Threat Models and Espionage Risks 14:38 AI and Cybersecurity: New Frontiers 19:41 Listener Questions and Community Engagement 22:53 The Evolution of AI Security Frameworks 26:29 Understanding New Attack Surfaces in AI 28:56 The Role of Automation in Security 31:05 Challenges of Non-Technical Users in Security 33:53 Best Practices for Managing Credentials 38:16 Building Security from the Ground Up 41:52 Compliance vs. Security in Startups 48:02 Understanding Security Constructs 51:06 Prioritizing Security Controls 52:48 The Role of SAST in Security 59:38 AI and Vulnerability Management 01:02:15 Coordinating Vulnerability Disclosure
S4 · E21
1 lug 2026
1:06:33
S4E20 - AI Executive Order, Project Glasswing Expanding, Cybersecurity Workforce
Promoting Advanced Artificial Intelligence Innovation and Security The White House EO pushes federal agencies toward AI-enabled cyber defense, frontier-model benchmarking, and a voluntary framework for trusted access to high-end AI systems. Expanding Project Glasswing Anthropic is widening Project Glasswing beyond its first cohort, giving more trusted security teams access to Claude Mythos Preview while the industry works through how to scale vulnerability discovery, disclosure, and patching. Securely testing on customer data The crew digs into the practical problem of validating AI and security tools against real customer environments without turning sensitive data into test exhaust, training material, or cross-tenant risk. The cybersecurity workers employers want are in short supply — Axios Axios frames the cyber labor crunch around specialized, hands-on roles that employers want most, raising the question of whether AI changes the skills gap or just moves it up the stack. Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, https://www.whirlai.com/) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E20
9 giu 2026
1:06:14
S4E19 – Canvas hacked, Cloudflare layoffs, GitHub CVE rundown
Canvas hack strands university students during finals week. A Canvas cyberattack hit universities and K-12 schools during finals, locking students and teachers out of grades, assignments, lecture materials, and exams at the worst possible moment. Building for the future. Cloudflare says it is cutting more than 1,100 employees as it restructures around internal AI-driven workflows, even as the timing alongside earnings and a sharp stock reaction raises harder questions about the story investors were told. GitHub RCE Vulnerability: CVE-2026-3854 Breakdown | Wiz Blog. Wiz breaks down a critical GitHub infrastructure flaw where an authenticated user could turn a normal git push into remote code execution on GitHub Enterprise Server, with GitHub.com mitigated and GHES customers urged to patch. Dirty Frag (CVE-2026-43284, CVE-2026-43500) Patches Released. AlmaLinux shipped kernel patches for Dirty Frag, a pair of Linux kernel bugs in IPsec ESP and rxrpc paths that can give local attackers root, with public exploit code already available. Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, stealth startup) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E19
12 mag 2026
1:09:12
S4E18 – Mythos and TPRM, does SOC 2 really work?
00:34 - Introduction 03:33 - Enterprise Challenges 07:08 - End User and Browsers 21:55 - Vulnerability Metrics 40:37 - Approaching Leadership 42:09 - TPRM Discussion 46:40 - Sharing Findings 01:03:04 - Conclusion Mozilla: Anthropic’s Mythos found 271 security vulnerabilities in Firefox 150 Anthropic’s Mythos found 271 zero-day vulnerabilities in Firefox 150 Mozilla let Anthropic’s Mythos loose on Firefox 150’s codebase, harvesting 271 shippable fixes in one sweep and forcing the security team to reckon with AI-scale fuzzing, triage, and patch velocity. https://arstechnica.com/ai/2026/04/mozilla-anthropics-mythos-found-271-zero-day-vulnerabilities-in-firefox-150/ Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, stealth startup) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E18
28 apr 2026
1:05:26
S4E17 – Mythos, Delve's downfall, and supply chain attacks
Project Glasswing (https://www.anthropic.com/glasswing) Anthropic is letting AWS, Apple, Google, Microsoft, JPMorgan, Cisco, NVIDIA, and friends point Claude Mythos at their shared attack surface while backing it with $100M in credits and $4M for OSS security groups so blue teams can burn down latent vulns before the offense gets equivalent AI. Inside the TeamPCP cascading supply chain attack (https://www.reversinglabs.com/blog/teampcp-supply-chain-attack-spreads) Hijacked Trivy GitHub Actions poisoned Docker images, stole CI secrets, and daisy-chained through Checkmarx workflows, npm packages, and VS Code extensions, seeding thousands of tenants with infostealers and proving CI creds are the new crown jewels. Delve – Fake Compliance as a Service - Part I (https://substack.com/home/post/p-191342187) A report says Delve mass-produced fake SOC 2 artifacts and funneled them through shell auditors, leaving customers—from indie apps to a Nasdaq firm—waving fraudulent attestations that crater their legal compliance. Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, stealth startup) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E17
23 apr 2026
1:08:58
Special RSAC episode with Cloudflare - Cybersecurity and AI, CISO/Board dynamics, future of cybersecurity
The Adversarial Podcast brings you a special episode in collaboration with Cloudflare's Security Signal Podcast. 0:39 - 3:33 AI Governance and Autonomy 6:26 - 8:49 Human in the Loop 9:17 - 11:40 Cybersecurity and AI 15:26 - 18:19 Resilience and Anti-Fragility 28:24 - 33:05 Threat Intelligence 33:31 - 36:50 Board and CISO Dynamics 41:09 - 42:35 Future of Cybersecurity 42:35 - 44:14 Books and Resources Security Signal Podcast: https://podcasts.apple.com/us/podcast/security-signal/id1815513800 Cloudflare; http://cloudflare.com/ Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, stealth startup) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E16
14 apr 2026
44:38
S4E15 – RSAC, Iranian hackers, White House's Cyber Strategy and Cyber EOs, the Future of TPRM
Iran-linked hackers claim responsibility for attack on US medical device maker Stryker Attackers tied to Iran say they hit Stryker, and investors punished the stock as the company scrambled to assess exposure. Trump Signs Executive Order Aimed at Cybercrime Gangs The President issued an order to tide together federal tools, international partners, and private-sector incentives for hunting down and disrupting ransomware crews. President Trump’s Cyber Strategy for America The new national cyber strategy leans hard on resilience, collaboration with allies, and deterring Beijing through offensive-ready posture. The future of third-party risk is NOT better questionnaires The author argues that automation and better data sharing—not more paperwork—are what finally move the needle on vendor risk management. Hosts: Jerry Perullo (Founder, https://adversarial.com/) Sounil Yu (Founder, https://www.knostic.ai/) Mario Duarte (Founder, stealth startup) Producer: Tillson Galloway (Founder, http://githoundexplore.com/)
S4 · E15
17 mar 2026
1:09:52
1 di 7