Modern Cyber with Jeremy Snyder

Modern Cyber with Jeremy Snyder

di Jeremy Snyder

Lavoro Tecnologia

Stagione 1

Mikko Hypponen of Sensofusion

In this kick-off episode for 2026, Jeremy is joined by the legendary Mikko Hypponen, Chief Research Officer at Sensofusion, for a comprehensive retrospective of 2025 and a look ahead at the future of AI-driven threats. Mikko, now a "Mount Rushmore" guest of the show, shares insights from his transition into the anti-drone space while reflecting on a year defined by massive infrastructure disruptions. The duo discusses the staggering impact of 2025 ransomware incidents, most notably the Jaguar Land Rover breach, which halted production for six weeks and cost an estimated £1.5 billion. Mikko argues that these events prove cybersecurity is no longer just about protecting computers—it’s about securing society itself. They also break down the "random shotgun" nature of modern attacks, where gangs like Clop and Akira target vulnerabilities rather than specific industries or geographies. Turning to AI, Mikko provides a reality check on the current state of deepfakes and automated orchestration. He reflects on the first massive AI-orchestrated cyber espionage campaign of 2025 and explains why the battle between open-source and closed-source models will define the next phase of defense. Finally, they examine how "data is the new oil" and AI is the "new oil refinery," creating a dual-extortion landscape where the risk of data leakage often outweighs the cost of downtime. About Mikko Mikko Hypponen is a world-renowned global security expert, author, and speaker with over 35 years of experience in the industry. In August 2025, Mikko transitioned from his long-standing tenure at WithSecure to become the Chief Research Officer at Sensofusion, a Finnish company specializing in advanced anti-drone technologies. Mikko has assisted law enforcement in the U.S., Europe, and Asia on major cybercrime cases since the 1990s and is the curator of the Malware Museum at the Internet Archive. He is the author of the best-selling book "If It's Smart, It's Vulnerable" and a frequent contributor to The New York Times, Wired, and Scientific American. In addition to his role at Sensofusion, Mikko serves as an advisor to Firetail. Episode Links https://sensofusion.com/ https://mikko.com/ https://www.firetail.ai/ai-breach-tracker https://www.anthropic.com/news/disrupting-AI-espionage

This Week in AI Security - 18th December 2025

In the final episode of 2025, Jeremy examines the evolution of SEO poisoning into "AI poisoning," a major privacy breach involving a popular browser extension, and shares a data-driven "sneak peek" at the state of AI security over the past year. Key Stories & Developments: AI Poisoning of Search Results: Researchers identified an attack where threat actors plant false information online to trick AI-powered search engine crawlers. This results in search engines providing AI summaries that list scam phone numbers for legitimate services like airline call centers, effectively creating a modern, AI-driven version of SEO poisoning. The "Pay-to-Crawl" Proposal: Jeremy discusses a new proposal from Creative Commons that suggests moving away from outright blocking AI crawlers. Instead, website owners could set a price for crawling and training, allowing organizations to monetize the use of their data by LLM providers. Urban VPN Privacy Breach: A popular Chrome and Edge extension, Urban VPN Proxy, was caught intercepting and reading the AI chat messages of its 7.3 million users. This incident highlights the risk of third-party browser extensions reading sensitive data that users assume is private. 2025 in Review Snapshot: Using data from the Firetail AI Incident Tracker, Jeremy reveals two major trends from 2025: The Surge in Incidents: AI security incidents saw a massive jump from 2024 to 2025, marking this as the year AI-related security became a global, pervasive problem. Disclosure vs. Injection: While the OWASp Top 10 lists prompt injection as the #1 risk, the tracker data shows that sensitive information disclosure (largely due to organizational error) actually outstrips prompt injection by about a third. Episode Links https://finance.yahoo.com/news/aurascape-researchers-expose-ai-attack-140000260.html?guccounter=1 https://techcrunch.com/2025/12/15/creative-commons-announces-tentative-support-for-ai-pay-to-crawl-systems/ https://thehackernews.com/2025/12/featured-chrome-browser-extension.html https://www.firetail.ai/ai-breach-tracker

Chris Parker of WhatIsMyIPAddress.com

In this episode of Modern Cyber, Jeremy is joined by Chris Parker, the founder of WhatIsMyIPAddress.com, one of the most visited websites in the world. With over 13 million monthly visitors, Chris has spent more than 25 years helping people understand their digital presence and protect their online privacy. The conversation dives into the fascinating 26-year history of the site—from its start as a simple hobby on a home Windows NT box to becoming a global authority on cybersecurity. Chris shares "war stories" from the early days of the web, including dealing with notoriously verbose log files that filled entire hard drives and managing a home data center that maxed out local copper lines. Chris and Jeremy also explore the modern landscape of digital privacy, discussing the balance between transparency and anonymity. They cover practical topics like how scammers use urgency to fleece victims, the "supply chain" risks of website plugins, and Chris's "middle-ground" approach to privacy—avoiding both complete exposure and the "Faraday cage" lifestyle. About Chris Parker Chris Parker is the founder of WhatIsMyIPAddress.com, one of the world’s most visited websites, helping more than 13 million people each month safeguard their digital privacy. Chris has become the go-to expert on protecting yourself in the digital age, whether from scammers, data miners, or privacy threats you didn't know existed. He is the author of Privacy Crisis: How to Maintain Your Privacy Without Becoming a Hermit, and host of The Easy Prey Podcast. Episode Links Website: https://www.privacycrisis.com LinkedIn: https://www.linkedin.com/in/christophergparker/ Podcast: https://www.easyprey.com/

This Week in AI Security - 11th December 2025

In this week's episode, Jeremy focuses on the escalating threat of prompt injection across the enterprise, the introduction of a new OWASP Top 10 list, and a surprising advisory from Gartner. Prompt Injection & RCE: PromptPwnd: A vulnerability in GitHub Actions allows attackers to use malicious commit messages to perform prompt injection against AI agents, executing privileged tools and leaking secrets from CI/CD pipelines. IDE Attack Surface: Similar prompt injection flaws were identified in popular development environments and extensions (Cursor, Copilot, Z-Ro), showing how malicious prompts can bypass guardrails and hijack context within the IDE. GeminiJack: A "zero-click" vulnerability in Google Gemini Enterprise and Vertex AI Search allowed attackers to embed indirect prompt injections in shared documents (Gmail, Calendar, Docs). A routine employee search would activate the attack, causing the AI to exfiltrate sensitive corporate data. Industry Shifts: Gartner's Advisory: Gartner issued an unusual strong advisory recommending that CISOs block all AI browsers (like ChatGPT Atlas and Perplexity Comet) for the foreseeable future due to inherent security risks, including data leakage, credential abuse, and autonomous rogue actions. New OWASp Top 10: The OWASp Top 10 for Agentic Applications was released, focusing on risks unique to autonomous, tool-using systems, such as Agent Goal Hijack, Identity and Privilege Abuse, and Agentic Supply Chain Vulnerabilities. Episode Links: https://gbhackers.com/prompt-injection-vulnerability-in-github-actions/ https://thehackernews.com/2025/12/researchers-uncover-30-flaws-in-ai.html https://securityboulevard.com/2025/12/indirect-malicious-prompt-technique-targets-google-gemini-enterprise/ https://securityboulevard.com/2025/12/gartners-ai-browser-ban-rearranging-deck-chairs-on-the-titanic/ https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/ ++++++++++ Worried about AI security? Get Complete AI Visibility in 15 Minutes. Discover all of your shadow AI now. Book a demo of Firetail's AI Security & Governance Platform: https://www.firetail.ai/request-a-demo

This Week in AI Security - 4th December 2025

In this week's episode, Jeremy dissects two critical security issues and shares key strategic takeaways from the recent Ascent Community Summit on Advancing AI Security. Security Incidents & Research: OpenAI Third-Party Breach: We examine the security incident where OpenAI was affected by a third-party breach via the Mixpanel analytics platform. While customer PII was exposed, prompt and data content was not impacted. Jeremy notes that the API was the attack surface, reinforcing a recurring theme in AI-related incidents. Adversarial Poetry: We break down a fascinating academic paper demonstrating that embedding malicious prompts inside poetry is a successful technique for bypassing LLM guardrails. In some models, this "adversarial poetry" increased the Attack Success Rate (ASR) by over 60%, showing how context manipulation can trick frontier models. Ascent Community Summit Takeaways: Jeremy shares high-level insights from the summit (co-hosted by Paladin and Georgia Tech), focusing on securing critical sectors (Defense, Infrastructure, Healthcare). Key themes include: Core Requirements for AI: The need for math expertise, dedicated compute infrastructure, massive data access, and specialized people. The New Perimeter: Discussion shifted from "identity as the perimeter" to data being the key asset and central focus for security controls. Supply Chain Risks: The societal impact of the AI boom, including increased strain on electricity, cooling, and bandwidth for data center infrastructure. Brakes on a Fast Car: The CISO's role is framed as enabling maximum speed while having the ability to act as the "brakes on a very fast moving car" (Dundee West, GSK), emphasizing rapid response over stagnation. Episode Links https://openai.com/index/mixpanel-incident/ https://arxiv.org/pdf/2511.15304 https://sites.gatech.edu/asccent/summit/ ------ Worried about AI security? Get Complete AI Visibility in 15 Minutes. Discover all of your shadow AI now. Book a demo of Firetail's AI Security & Governance Platform: https://www.firetail.ai/request-a-demo

This Week in AI Security - 27th November 2025

In this week's episode, Jeremy covers seven stories that highlight the continuing pattern of API-level risks, the rise of multi-agent threats, and new academic insights into LLM fundamentals. Key stories include: RCE via PyTorch: A high-severity vulnerability (with an assigned CVE) was discovered in the widely-used PyTorch package, enabling Remote Code Execution (RCE) through malicious payloads at the API layer. This reinforces the trend of the API being the primary attack surface for AI applications. AI Browser Local Command Execution: Researchers found an API flaw in AI browsers that allowed a malicious instruction set to execute local commands on a user's machine via an embedded extension. Klein Bot Vulnerabilities: An open-source coding agent was found to have multiple security flaws, including the exfiltration of API keys and the disclosure of its underlying model (Grok), validating OWASp's risk categories. Multi-Agent Risk in ServiceNow: Researchers demonstrated that in ServiceNow’s new A-to-A agentic workflows, default configurations place agents in the same network, allowing them to communicate and be exploited using the privileges of the human user who created them. The "Subspace Problem" of Red Teaming: Academic research argues that current LLM red teaming methods are flawed because they test human language, not the numerical token strings the LLM actually processes, meaning predictable token-level vulnerabilities remain hidden. AI Evaluation Shift: A paper argues that non-deterministic LLM environments require a shift away from binary "yes/no" security checks (like traditional network security) toward scenario-based testing for better risk evaluation. Positive ROI of AI in Security: A Google paper provides positive data for early movers, showing that AI can triage at least 50% of security incidents, leading to reduced human workloads and faster response times, providing a strong case for simple, prompt-based AI improvements in security operations. ------ Worried about AI security? Get Complete AI Visibility in 15 Minutes. Discover all of your shadow AI now. Book a demo of Firetail's AI Security & Governance Platform: https://www.firetail.ai/request-a-demo

Adam Pilton of Heimdal

In this episode of Modern Cyber, Jeremy is joined by Adam Pilton, a cybersecurity expert with a background of 15 years in law enforcement, where his final role was as a Detective Sergeant leading the Covert Operations and Cybercrimes team. Drawing on his unique experience investigating and prosecuting hundreds of offenders, Adam provides a frontline perspective on the current state of cybercrime, noting that cybercriminals are "getting better and stronger" while individuals and businesses are "not keeping up". The conversation focuses on the human and organizational challenges in cybersecurity, stressing that small businesses should abandon the belief that they are too small to be targeted, as attackers "hit small businesses all day long" for incremental profit. Adam discusses the severe practical impacts of attacks, warning that businesses must "expect downtime" and be prepared for the significant time needed for recovery. He advocates for storytelling and analogies (like the comparison of hacking to a burglary) over technical regulations to build a strong security culture. Adam also shares insights from his post-law enforcement work as an auditor and consultant, highlighting the common organizational "motivation problem" where people acknowledge the risk but delay action, comparing it to perpetually starting a diet "tomorrow". Finally, he addresses the breakdown of trust in the age of deepfakes (citing the Irish election example) and the critical need for continuous tabletop exercises to test communication and expose "little gaps" before a crisis hits. Guest Bio – Adam Pilton With a background of 15 years in law enforcement, Adam's final role was as a Detective Sergeant leading the Covert operations and Cyber Crime teams. Since then, Adam has worked in cyber security since 2016 across various roles and has a broad understanding of cyber security, from the impact of cyber crime upon individuals and businesses to the need to convey the right messages to senior leaders and end users, ensuring engagement and support. As a subject matter expert in multiple areas for a large organisation, Adam has investigated and supervised hundreds of cases, identifying and prosecuting offenders. He has introduced digital tactics into overt and covert investigations, developing digital capabilities. Adam also held responsibility for training, utilising his communication skills to simplify the complex. Adam has worked with multi-national businesses developing their people and processes to improve their cyber security maturity. Episode Links https://heimdalsecurity.com/ https://www.linkedin.com/in/adampilton/

This Week in AI Security - 20th November 2025

In this week's episode, Jeremy covers two major and critical developments that underscore the need to harden the foundational components of AI systems and recognize the reality of AI-orchestrated attacks. First, we analyze Shadow MQ, a vulnerability discovered by Oligo that affects multiple popular AI tools, including those from Nvidia and Meta Llama. The flaw stems from the mass reuse of core, insecure components—specifically, an unsafe Python pickle deserialization technique—in the underlying plumbing of various LLMs. This vulnerability allows attackers to inject malicious commands, potentially leading to Remote Code Execution (RCE) and Privilege Escalation at the API layer. Second, we dive deep into the first publicly confirmed, AI-orchestrated cyber espionage campaign, detailed in a threat intelligence report from Anthropic. The state-sponsored campaign used a frontier AI model to accelerate nearly every phase of the attack, including: Weaponized System Prompts: Attackers defined a persona ("senior cyber operations specialist") to guide the LLM's malicious behavior. AI-Driven Evasion: The AI was used to refine malware and bypass EDR solutions. AI-Powered Reconnaissance: The model performed vulnerability research on obscure protocols and orchestrated lateral movement within networks. Jeremy emphasizes that this report is a wake-up call, validating the core risks around AI adoption and proving that malicious AI usage is now a real-world reality. Episode Links: https://www.oligo.security/blog/shadowmq-how-code-reuse-spread-critical-vulnerabilities-across-the-ai-ecosystem https://assets.anthropic.com/m/ec212e6566a0d47/original/Disrupting-the-first-reported-AI-orchestrated-cyber-espionage-campaign.pdf ------ Worried about AI security? Get Complete AI Visibility in 15 Minutes. Discover all of your shadow AI now. Book a demo of Firetail's AI Security & Governance Platform: https://www.firetail.ai/request-a-demo

Ben Wilcox of ProArch

In this episode of Modern Cyber, Jeremy is joined by Ben Wilcox, the unique combination of CTO and CISO at ProArch, to discuss navigating the critical intersection of speed, risk, and security in the era of AI. Ben shares his perspective as a long-time practitioner in the Microsoft ecosystem, emphasizing that the security stack must evolve with each major technology shift—from on-prem to cloud to AI. The conversation focuses on how to help customers achieve "data readiness" for AI adoption, particularly stressing that organizational discipline (like good compliance) is the fastest path to realizing AI's ROI. Ben reveals that the biggest concern he hears from enterprise customers is not LLM hallucinations or bias, but the risk of a major data breach via new AI services. He explains how ProArch leverages the comprehensive Microsoft security platform to provide centralized security and identity control across data, devices, and AI agents, ensuring that user access and data governance (Purview) trickle down through the entire stack. Finally, Ben discusses the inherent friction of his dual CISO/CTO role, explaining his philosophy of balancing rapid feature deployment with risk management by defining a secure "MVP" baseline and incrementally layering on controls as product maturity and risk increase. About Ben Wilcox Ben Wilcox is the Chief Technology Officer and Chief Information Security Officer at ProArch, where he leads global strategy for cloud modernization, cybersecurity, and AI enablement. With over two decades of experience architecting secure digital transformations, Ben helps enterprises innovate responsibly while maintaining compliance and resilience. He’s recently guided Fortune 500 clients through AI adoption and zero-trust initiatives, ensuring that security evolves in step with rapid technological change. Episode Links https://www.proarch.com/ https://www.linkedin.com/in/ben-wilcox/ https://ignite.microsoft.com/en-US/home

This Week in AI Security - 13th November 2025

In this week's episode, Jeremy covers seven significant stories and academic findings that reveal the escalating risks and new attack methods targeting Large Language Models (LLMs) and the broader AI ecosystem. Key stories include: Prompt Flux Malware: Google Threat Intelligence Group (GTAG) discovered a new malware family called Prompt Flux that uses the Google Gemini API to continuously rewrite and modify its own behavior to evade detection—a major evolution in malware capabilities. ChatGPT Leak: User interactions and conversations with ChatGPT have been observed leaking into Google Analytics and the Google Search Console on third-party websites, potentially exposing the context of user queries. Traffic Analysis Leaks: New research demonstrates that observers can deduce the topics of a conversation in an LLM chatbot with high accuracy simply by analyzing the size and frequency of encrypted network packets (token volume), even without decrypting the data. Secret Sprawl: An analysis by Wiz found that several of the world's largest AI companies are leaking secrets and credentials in their public GitHub repositories, underscoring that the speed of AI development is leading to basic, repeatable security mistakes. Non-Deterministic LLMs: Research from Anthropic highlights that LLMs are non-deterministic and highly unreliable in describing their own internal reasoning processes, giving inconsistent responses even to minor prompt variations. The New AI VSS: The OWASp Foundation unveiled the AI Vulnerability Scoring System (AI VSS), a new framework to consistently classify and quantify the severity (on a 0-10 scale) of risks like prompt injection in LLMs, helping organizations make better risk-informed decisions. Episode Links: https://cybersecuritynews.com/promptflux-malware-using-gemini-api/ https://thehackernews.com/2025/11/microsoft-uncovers-whisper-leak-attack.html https://arstechnica.com/ai/2025/11/llms-show-a-highly-unreliable-capacity-to-describe-their-own-internal-processes/ https://futurism.com/artificial-intelligence/llm-robot-vacuum-existential-crisis https://www.scworld.com/resource/owasp-global-appsec-new-ai-vulnerability-scoring-system-unveiled https://arstechnica.com/tech-policy/2025/11/oddest-chatgpt-leaks-yet-cringey-chat-logs-found-in-google-analytics-tool/ https://www.securityweek.com/many-forbes-ai-50-companies-leak-secrets-on-github/

6 di 14