AI Security Update

In this episode, we sit down with Jacob Krell, a senior director for Secure AI solutions and cybersecurity at Suzu Labs. Jacob discusses how AI agents are changing offensive security: they can automate multi-step attacks, adapt quickly, and increase the potential impact of a single attacker. He explains why security teams need to focus on permissions, identity checks, guardrails, human oversight, and detailed logging—not just model capabilities. The conversation also touches on how to break into an offensive cybersecurity career through hands-on practice, certifications, labs, and curiosity.

*** For educational purposes only ***

DISCLAIMER: This video is for educational and informational purposes only.

The content is intended to demonstrate concepts, techniques, and processes for learning and research purposes. It is not intended to encourage, promote, or facilitate any illegal activity, cyberattack, or unauthorized access to systems.

By watching this video, you agree that you will not use the information to commit any unlawful act. The creator of this content makes no warranty, express or implied, and shall not be liable for any damages arising from the use of the information provided.

This video does not constitute professional advice, and you should consult qualified professionals before applying any techniques shown in a real‑world context.

AI attacks are getting faster, smarter, and harder to stop, and most security teams are still built for a world that no longer exists. Gina Yacone, Regional and Advisory CISO at Trace3, explains why the real threat is not just malicious code, but autonomous, self-healing attacks that move faster than traditional defenses can respond.

We talk about the new Anthropic report "Detecting and countering misuse of AI" that was released in September 2026

Gina breaks down new AI attack paths, supply chain risk, vendor pivots, runtime security, and the growing blast radius of autonomous threats. She also explains why having a “human in the loop” is not enough—the real question is where humans should sit in the process so they can provide judgment without becoming bottlenecks.

This is an essential conversation for security professionals, enterprise leaders, and anyone who wants to understand what AI-enabled attacks mean for their organization, data, and family.

Link to Anthropic report: https://www.anthropic.com/threat-intelligence-report-september-2026

In this episode, Fredrick Wahome, cybersecurity expert and Vice President of the Kenya Cyber Security and Forensics Association, joins the show from Nairobi to explore how AI is reshaping the security landscape across Africa and beyond.

We dive into AI-assisted ransomware, ransomware-as-a-service, incident response, offline backups, open-source security risks, threat intelligence, and the growing need for AI-specific security assessments. Fredrick also explains why cybersecurity awareness is struggling to keep pace with AI adoption and what organizations, technical teams, families, and everyday users can do to stay protected.

The conversation also explores AI governance and cybersecurity policy across Africa, including the need for security by design and stronger alignment with emerging standards such as ISO 42001.

AI agents are moving fast, and most organizations are not ready for what happens when they get a key, a task, and access. John Morton reveals why the real AI security risk is not just the model - it is the identity, permissions, and guardrails around the digital worker itself. We discuss the recent attacks on water treatment plants as a case study and considerations of using AI in an educational setting.

John brings 30+ years in cyber, infrastructure, cloud, and digital transformation, plus his perspective as an adjunct faculty member in George Washington University's cyber analytics program. The conversation goes beyond enterprise risk. John and Anmol get into the growing threat to critical infrastructure, including water systems, power grids, and other legacy environments with technical debt and huge attack surfaces. John lays out what state and local leaders need to do first: understand what they have, map the interconnections, modernize where needed, and make AI security a legislative priority instead of an afterthought.

If you care about AI security, cyber policy, critical infrastructure, or the future of education in an AI-driven world, this episode is packed with practical guidance. Essential listening if you want to understand how to keep moving fast with AI without handing over the master key.

The real risk to AI lives in the ecosystem - chips, minerals, data centers, talent, geopolitics, and the criminals already exploiting all of it.

Dr. Cuneyt Gurer breaks down why security teams that only focus on the model are missing the bigger picture, and why AI is now colliding with illicit markets, organized crime, and state proxies. Anmol and Dr. Gurer unpack how AI changes the nature of traditional threats. You’ll discover why supply chains matter just as much as prompts, and how sovereign AI is tied to minerals and chip production

Dr. Gurer is a professor of international and security studies at the George C. Marshall European Center for Security Studies, where he designs courses for security professionals and studies how AI intersects with geopolitics, illicit markets, and defense. The opinions reflected in this episode are his own personal views and do not reflect those of any organization.

Emma Fang, senior security architect at EPAM and vice president at Women in Cybersecurity's UK and Ireland affiliate, draws from a decade of cybersecurity experience both as a practitioner and as a hiring manager for AI security roles. Her insights are vital for defenders and aspiring AI security professionals alike.

We talk about the incident where OpenAI broke out of its test environment, impacting Hugging Face, the changing landscape of AI attacks, and tips and tricks on how to get hired in AI security jobs.

Don't miss Emma's upcoming talks at DEF CON’s AI and AppSec villages, where she explores the failures of traditional zero trust and the risks of shadow APIs

In this episode, software veteran Carlos Echeverria, founder of QinniLabs — with over 20 years of experience at Fortune100 corporations including Amazon, Wells Fargo, and Allstate — uncovers the real-world strategies for managing AI risks in engineering. From implementing Behavior Driven Development (BDD) to enforcing non-negotiable tests, he shares how structured workflows can prevent ghost decisions, greenwashing, and uncontrolled AI chaos.

Discover how agile methods like Scrum and specific guardrails in agents.md can transform your AI team into a predictable, secure powerhouse. Carlos emphasizes that integrating traditional engineering principles—such as specification-driven development and rigorous governance—is more critical than ever in the age of rapid AI adoption. You’ll learn how to inject discipline into your AI workflows, avoid hallucinations, and maintain human oversight in a landscape pushing towards autonomous decision-making.

In this episode, we sit down with Mert Can Boyar, a privacy lawyer based in Istanbul, Turkey blending law, technology, and playful education. He also founded the privacy village which was featured in Forbes. Discover how creative approaches like games, musicals, and comic books are revolutionizing privacy and cybersecurity awareness, especially in complex areas like AI ethics and data privacy.

In this episode, you’ll learn how privacy-enhancing technologies such as homomorphic encryption and differential privacy are shaping privacy-preserving AI models. See how innovative educational tools like gamified festivals and storytelling are making complex AI and security concepts accessible and engaging for everyone, from legal teams to everyday users.

We break down the evolving landscape of AI regulation, highlighting differences between the EU, US, and Turkey. Understand how the EU’s risk-based approach and the AI Act could impact developers and companies worldwide.

This episode is essential listening for legal professionals, security experts, AI developers, and anyone interested in safeguarding the future of AI. Learn how to transform "boring" privacy and security topics into engaging, impactful experiences via games, stories, and comics.

What happens when AI agents start behaving like autonomous users inside your infrastructure? In this episode, we sit down with Yacine Souam, an R&D engineer at Escape — the Y Combinator-funded, Forbes 30 Under 30-recognized cybersecurity startup — to explore one of the fastest-growing frontiers in tech: the intersection of AI security and API security.

Yacine shares his journey through cybersecurity and AI research, along with his work on the OWASP Exchange, an initiative helping bridge the gap between security professionals, developers, and AI practitioners. The conversation dives into why AI security is becoming inseparable from API security, especially as AI agents increasingly interact with internal systems, APIs, and sensitive data.

We explore the emerging challenges companies face in securing AI models and autonomous agents, from observability gaps and permission scoping to adversarial machine learning and prompt injection attacks. Yacine explains why AI agents should be treated as non-human identities and how principles like least privilege and security-by-design are becoming essential for modern AI deployments.

The episode also covers real-world security concerns, including supply chain risks in the AI ecosystem and lessons from the recent LiteLLM incident. Along the way, Yacine shares practical advice for engineers, founders, and security teams looking to safely adopt AI while maintaining innovation speed.

Whether you’re building AI products, securing APIs, or simply trying to understand where the future of cybersecurity is headed, this episode offers a grounded and highly practical look into the rapidly evolving world of AI security.

Stay curious, keep experimenting, and learn how security is evolving alongside AI.

This episode will change what you think about Meta smart glasses, AI governance, and compliance with accessibility requirements. In this episode, we are joined by Tiffani Martin, founder of Visio Tech, to talk about accessibility for Global Accessibility Awareness Day. We talk about the helpful benefits as well as security and privacy considerations for Meta glasses, smart canes, and assistive technology. We also cover why accessibility needs to be considered alongside the traditional pillars of AI governance.