How AI Will Transform Society and Affect the Cybersecurity Field

Summary: Timothy De Block sits down with Matt Topper of Uber Ether to discuss the critical intersection of Identity and Access Management (IAM) and the current cyber threat landscape. They explore how adversaries have shifted their focus to compromising user accounts and non-human identities, making identity the "last threat of security". Matt Topper argues that most enterprise Zero Trust implementations are merely "VPN 2.0" and fail to integrate the holistic signals needed for true protection. The conversation dives into the rise of cybercrime as a full-fledged business, the challenges of social engineering, and the promising future of frameworks like Shared Signals to fight back. Key Takeaways The Identity Crisis in Cybersecurity The Easiest Way In: With security tooling improving, attackers focus on compromising user accounts or stealing OAuth tokens and API keys to gain legitimate access and exfiltrate data. Cybercrime as a Business: Cybercriminal groups now operate like legitimate businesses, with HR, marketing, and executives, selling initial access and internal recon capabilities to other groups for a cut of the final ransom. The Insider Threat: Cybercriminals are increasingly paying disgruntled employees for their corporate credentials, sometimes offering a percentage of the final ransom (which can be millions of dollars) or just a few thousand dollars. Social Engineering the Help Desk: Attackers easily bypass knowledge-based authentication (KBA) questions because personal data has been leaked and they exploit the help desk's desire to be helpful under pressure to gain access. Zero Trust, Non-Human Identity, and the Path Forward Zero Trust is Underwhelming: Matt Topper views most enterprise implementations of Zero Trust as overly network-centric "VPN 2.0" that fail to solve problems for multi-cloud or SaaS-based organizations. True Zero Trust is a holistic strategy that requires linking user, device, and machine-to-machine signals. The Non-Human Identity Problem: Organizations must focus on mapping and securing non-human identities, which include API keys, service accounts, servers, mobile devices, and runners in CI/CD pipelines. These keys often have broad access and are running unchecked. Shared Signals Framework (SSF): A promising solution developed by the OpenID Foundation, SSF allows large vendors (like Microsoft, Google, and Salesforce) to share risk and identity signals. This allows a company to automatically revoke a user's session in a third-party application if a compromise is detected by the identity provider. User Behavior Analytics (UBA): Effective security requires UBA, such as tracking users' browsing habits and using data analytics to establish a baseline of normal behavior, moving toward the "Moneyball" approach seen in sports. Data Quality and the IAM Challenge Data Quality is Broken: Many problems in IAM stem from poor data quality in source systems like HR and Active Directory, where there is no standardization, legacy data remains, and roles are misaligned. Selling Security to Marketing: To gain funding and traction for UBA and data analytics, security teams should pitch the problem to the marketing team by showing how it can track user behavior, prevent fraud (like "pizza hacks" from rewards program abuse), and save the company money in chargebacks. Resources & Contact UberEther: Matt Topper's company, which focuses on integrating identity access management tools to build secure systems right from day one. Shared Signals Framework (SSF): A framework from the OpenID Foundation for sharing security and identity signals across vendors.

Summary: Timothy De Block speaks with Bruce Johnson of TekStream about a truly innovative solution to the cybersecurity skills shortage: the Student-Powered Security Operations Center (SOC). Bruce outlines how this three-way public-private partnership not only provides 24-hour threat detection and remediation serves as a robust workforce development program for university students. The conversation dives into the program's unique structure, its 100% placement rate for students, the challenges of AI "hallucinations", and how the program teaches crucial life skills like accountability and critical thinking. The Student-Powered SOC Model Workforce Development: The program tackles the cybersecurity skills shortage by providing students with practical, real-world experience and helps bridge the gap where new graduates struggle to find jobs due to minimum experience requirements. Funding Structure: The program is built on a three-way private-public partnership involving the state, educational institutions, and Techstream. The funding for the SOC platform is often separate from the academic funding for student talent building. "Investment Solution": The model is positioned as an investment rather than an outsourced expense. Institutions own the licenses for their SIM environments and retain built assets, fostering collaborative value building. Reputational Value: The program provides significant reputational value to schools, boasting a 100% placement rate for students and differentiating them from institutions that only offer academic backgrounds. Cost Savings: It serves as a cost-saving measure for CISOs, as students are paid an hourly rate to perform security analyst work. Student Training and Impact Onboarding and Assessment: The formal onboarding process, which includes training on tools, runbooks, and hands-on labs, has been shortened to six weeks. The biggest indicator of a student's success is their critical thinking test, which assesses logical reasoning rather than rote knowledge. Progression and Mentorship: Students are incrementally matured by starting with low-complexity threats (like IP reputation) and gradually advancing to higher-difficulty topics, including TTPs (Tactics, Techniques, and Procedures), utilizing a complexity scoring system. Integrated career counseling meets regularly with students to review their metrics and guide their career planning. Metrics and Productivity: The program has proven successful, with students handling 50% of incident volume within a quarter of onboarding, including medium to high complexity threats. Beyond Cybersecurity: Students gain valuable, transferable life skills, such as collaboration, accountability, professionalism, and "adulting", which helps isolated students become more engaged. AI and the "Expert in the Loop" Techstream’s Overkill AI: Techstream uses its product, Overkill, for 24-hour threat detection and remediation, automating analysis, prioritization, and the creation of new detections to go "from zero to hero in 24 hours". Expert Supervision: Their approach is "expert in the loop" , meaning humans (students and analysts) are involved in supervising the AI, with automation being adopted incrementally as trust is built. The Hallucination Challenge: Timothy De Block raised concern about students lacking the experience to discern incorrect information or "hallucinations" from AI output. Bruce Johnson affirmed that the program trains students in three areas: using AI, supervising AI, and understanding AI broadly. Training Necessity: Students must learn how to do the traditional level one work before they can effectively supervise an AI, as experience is needed to detect when the AI makes a bad assumption.

Summary: Timothy De Block is joined by Sam Chehab to unpack the key findings of the 2025 Postman State of the API Report. Sam emphasizes that APIs are the connective tissue of the modern world and that the biggest security challenges are rooted in fundamentals. The conversation dives deep into how AI agents are transforming API development and consumption, introducing new threats like "rug pulls" , and demanding higher quality documentation and error messages. Sam also shares actionable advice for engineers, including a "cheat code" for getting organizational buy-in for AI tools and a detailed breakdown of the new Model Context Protocol (MCP). Key Insights from the State of the API Report API Fundamentals are Still the Problem: The start of every security journey is an inventory problem (the first two CIS controls). Security success is a byproduct of solving collaboration problems for developers first. The Collaboration Crisis: 93% of teams are struggling with API collaboration, leading to duplicated work and an ever-widening attack surface due to decentralized documentation (Slack, Confluence, etc.). API Documentation is Up: A positive sign of progress is that 58% of teams surveyed are actively documenting their APIs to improve collaboration. Unauthorized Access Risk: 51% of developers cite unauthorized agent access as a top security risk. Sam suspects this is predominantly due to the industry-wide "hot mess" of secrets management and leaked API keys. Credential Amplification: This term is used to describe how risk is exponential, not linear, when one credential gains access to a service that, in turn, has access to multiple other services (i.e., lateral movement). AI, MCP, and New Security Challenges Model Context Protocol (MCP): MCP is a protocol layer that sits on top of existing RESTful services, allowing users to generically interact with APIs using natural language. It acts as an abstraction layer, translating natural language requests into the proper API calls. The AI API Readiness Checklist: For APIs to be effective for AI agents: Rich Documentation: AI thrives on documentation, which developers generally hate writing. Using AI to write documentation is key. Rich Errors: APIs need contextual error messages (e.g., "invalid parameter, expected X, received Y") instead of generic messages like "something broke". AI Introduces Supply Chain Threats: The "rug pull" threat involves blindly trusting an MCP server that is then swapped out for a malicious one. This is a classic supply chain problem (similar to NPM issues) that can happen much faster in the AI world. MCP Supply Chain Risk: Because you can use other people's MCP servers, developers must validate which MCP servers they're using to avoid running untrusted code. The first reported MCP hack involved a server that silently BCC'd an email to the attacker every time an action was performed. Actionable Advice and Engineer "Cheat Codes" Security Shift-Left with Postman: Security teams should support engineering's use of tools like Postman because it allows developers to run security tests (load testing, denial of service simulation, black box testing) themselves within their normal workflow, accelerating development velocity. API Key Management is Critical: Organizations need policies around API key generation, expiration, and revocation. Postman actively scans public repos (like GitHub) for leaked Postman keys, auto-revokes them, and notifies the administrator. Getting AI Buy-in (The Cheat Code): To get an AI tool (like a Postman agent or a code generator) approved within your organization, use this tactic: Generate a DPA (Data Processing Agreement) using an AI tool. Present the DPA and a request for an Enterprise License to Legal, Security, and your manager. This demonstrates due diligence and opens the door for safe, approved AI use, making you an engineering "hero". About Postman and the Report Postman's Reach: Postman is considered the de facto standard for API development and is used in 98% of the Fortune 500. Report Origins: The annual report, now in its seventh year, was started because no one else was effectively collecting and synthesizing data across executives, managers, developers, and consultants regarding API production and consumption.

Summary: Timothy De Block sits down with Ed Gaudet, CEO of Censinet and a fellow podcaster, for a wide-ranging conversation on the rapid, transformative impact of Artificial Intelligence (AI). Ed Gaudet characterizes AI as a fast-moving "hammer" that will drastically increase productivity and reshape the job market, potentially eliminating junior software development roles. The discussion also covers the societal risks of AI, the dangerous draw of "digital cocaine" (social media), and Censinet's essential role in managing complex cyber and supply chain risks for healthcare organizations. Key Takeaways AI's Transformative & Disruptive Force A Rapid Wave: Ed Gaudet describes the adoption of AI, particularly chat functionalities, as a rapid, transformative wave, surpassing the speed of the internet and cloud adoption due to its instant accessibility. Productivity Gains: AI promises immense productivity, with the potential for tasks requiring 100 people and a year to be completed by just three people in a month. The Job Market Shift: AI is expected to eliminate junior software development roles by abstracting complexity. This raises concerns about a future developer shortage as senior architects retire without an adequate pipeline of talent. Adaptation, Not Doom: While acknowledging significant risks, Ed Gaudet maintains that humanity will adapt to AI as a tool—a "hammer"—that will enhance cognitive capacity and productivity, rather than making people "dumber". The Double-Edged Sword: Concerns exist over the nefarious uses of AI, such as deepfakes being used for fraudulent job applications, underscoring the ongoing struggle between good and evil in technology. Cyber Risk in Healthcare and Patient Safety Cyber Safety is Patient Safety: Due to technology's deep integration into healthcare processes, cyber safety is now directly linked to patient safety. Real-World Consequences: Examples of cyber attacks resulting in canceled procedures and diverted ambulances illustrate the tangible threat to human life. Censinet's Role: Censinet helps healthcare systems manage third-party, enterprise cyber, and supply chain risks at scale, focusing on proactively addressing future threats rather than past ones. Patient Advocacy: AI concierge services have the potential to boost patient engagement, enabling individuals to become stronger advocates for their own health through accessible second opinions. Technology's Impact on Mental Health & Life "Digital Cocaine": Ed Gaudet likened excessive phone and social media use, particularly among younger generations, to "digital cocaine"—offering short-term highs but lacking nutritional value and promoting technological dependence. Life-Changing Tools: Ed Gaudet shared a powerful personal story of overcoming alcoholism with the help of the Reframe app, emphasizing that the right technology, used responsibly, can have a profound, life-changing impact on solving mental health issues. Resources & Links Mentioned Censinet: Ed Gaudet's company, specializing in third-party and enterprise risk management for healthcare. Reframe App: An application Ed Gaudet used for his personal journey of recovery from alcoholism, highlighting the power of technology for mental health.

Wes (@kai5263499) spoke about this topic at BSides Hunstville this year. I was fascinated by it and decided to invite Wes on. Mac malware is a bit of an interest for Wes. He's done a lot of research on it. His talk walks through the history of malware on Macs. For Apple fan boys, Macs are still one of the more safer options in the personal computer market. That is changing though. Macs because of their increased market share are getting targeted more and more. We discuss some pretty nifty tools that will help with fending off that nasty malware. Little Snitch is one of those tools. Some malware actively avoids the application. Tune in for some more useful information.

Claire (@ClaireTills) doesn’t have your typical roll in infosec. She sits between the security teams and marketing team. It’s a fascinating roll and something that gives her a lot of insight into multiple parts of the business. What works and what doesn’t work in communicating security to the different areas. Check her blog out.In this episode we discuss:How important is it for the company to take security seriouslyHow would someone get started improving communication?Why we have a communication problem in infosecWhere should people startMore resources:Networking with Humans to Create a Culture of Security by Tracy Maleeff - BSides NoVa 2017Courtney K BsidesLV 2018, Implementing the Three Cs of Courtesy, Clarity, and Comprehension to Optimize End User Engagement (video not available yet)BSidesWLG 2017 - Katie Ledoux - Communication: An underrated tool in the infosec revolutionJeff Man, The Art of the Jedi Mind TrickThe Thing Explainer: Complicated Stuff in Simple WordsChris Roberts, Communication Across Ranges

Claire (@ClaireTills) doesn’t have your typical roll in infosec. She sits between the security teams and marketing team at Tenable. It’s a fascinating roll and something that gives her a lot of insight into multiple parts of the business. What works and what doesn’t work in communicating security to the different areas. Check her blog out.In this episode we discuss:What Claire’s experience is with communication and infosecWhat’s ahead for communication in infosecWhy do people do what they do?What questions to askMore resources:Networking with Humans to Create a Culture of Security by Tracy Maleeff - BSides NoVa 2017Courtney K BsidesLV 2018, Implementing the Three Cs of Courtesy, Clarity, and Comprehension to Optimize End User Engagement (video not available yet)BSidesWLG 2017 - Katie Ledoux - Communication: An underrated tool in the infosec revolutionJeff Man, The Art of the Jedi Mind TrickThe Thing Explainer: Complicated Stuff in Simple WordsChris Roberts, Communication Across Ranges

Summary: Timothy De Block is joined by Keith Hoodlet, Engineering Director at Trail of Bits, for a fascinating, in-depth look at AI red teaming and the security challenges posed by Large Language Models (LLMs). They discuss how prompt injection is effectively a new form of social engineering against machines, exploiting the training data's inherent human biases and logical flaws. Keith breaks down the mechanics of LLM inference, the rise of middleware for AI security, and cutting-edge attacks using everything from emojis and bad grammar to weaponized image scaling. The episode stresses that the fundamental solutions—logging, monitoring, and robust security design—are simply timeless principles being applied to a terrifyingly fast-moving frontier. Key Takeaways The Prompt Injection Threat Social Engineering the AI: Prompt injection works by exploiting the LLM's vast training data, which includes all of human history in digital format, including movies and fiction. Attackers use techniques that mirror social engineering to trick the model into doing something it's not supposed to, such as a customer service chatbot issuing an unauthorized refund. Business Logic Flaws: Successful prompt injections are often tied to business logic flaws or a lack of proper checks and guardrails, similar to vulnerabilities seen in traditional applications and APIs. Novel Attack Vectors: Attackers are finding creative ways to bypass guardrails: Image Scaling: Trail of Bits discovered how to weaponize image scaling to hide prompt injections within images that appear benign to the user, but which pop out as visible text to the model when downscaled for inference. Invisible Text: Attacks can use white text, zero-width characters (which don't show up when displayed or highlighted), or Unicode character smuggling in emails or prompts to covertly inject instructions. Syntax & Emojis: Research has shown that bad grammar, run-on sentences, or even a simple sequence of emojis can successfully trigger prompt injections or jailbreaks. Defense and Design LLM Security is API Security: Since LLMs rely on APIs for their "tool access" and to perform actions (like sending an email or issuing a refund), security comes down to the same principles used for APIs: proper authorization, access control, and eliminating misconfiguration. The Middleware Layer: Some companies are using middleware that sits between their application and the Frontier LLMs (like GPT or Claude) to handle system prompting, guard-railing, and filtering prompts, effectively acting as a Web Application Firewall (WAF) for LLM API calls. Security Design Patterns: To defend against prompt injection, security design patterns are key: Action-Selector Pattern: Instead of a text field, users click on pre-defined buttons that limit the model to a very specific set of safe actions. Code-Then-Execute Pattern (CaMeL): The first LLM is used to write code (e.g., Pythonic code) based on the natural language prompt, and a second, quarantined LLM executes that safer code. Map-Reduce Pattern: The prompt is broken into smaller chunks, processed, and then passed to another model, making it harder for a prompt injection to be maintained across the process. Timeless Hygiene: The most critical defenses are logging, monitoring, and alerting. You must log prompts and outputs and monitor for abnormal behavior, such as a user suddenly querying a database thousands of times a minute or asking a chatbot to write Python code. Resources & Links Mentioned Trail of Bits Research: Blog: blog.trailofbits.com Company Site: trailofbits.com Weaponizing image scaling against production AI systems Call Me A Jerk: Persuading AI to Comply with Objectionable Requests Securing LLM Agents Paper: Design Patterns for Securing LLM Agents against Prompt Injections. Camel Prompt Injection Defending LLM applications against Unicode character smuggling Logit-Gap Steering: Efficient Short-Suffix Jailbreaks for Aligned Large Language Models LLM Explanation: Three Blue One Brown (3Blue1Brown) has a great short video explaining how Large Language Models work. Lakera Gandalf: Game for learning how to use prompt injection against AI Keith Hoodlet's Personal Sites: Website: securing.dev and thought.dev

Summary: Join Timothy De Block for a special, behind-the-scenes episode where he rehearses his presentation, "The Hitchhiker's Guide to Threat Modeling." This episode serves as a unique guide for aspiring and experienced speakers, offering a candid look at the entire preparation process—from timing and slide design to audience engagement and controlled chaos. In addition to public speaking tips, Timothy provides a concise and practical overview of threat modeling, using real-world examples to illustrate its value. Key Presentation Tips & Tricks Practice for Time: Practice the presentation multiple times to ensure the pacing is right. Timothy suggests aiming to be a little longer than the allotted time during practice, as adrenaline and nerves on the day of the talk will often cause a person to speak more quickly. Use Visuals Strategically: Pacing and hand gestures can improve the flow of a talk. Be careful with distracting visuals, such as GIFs, by not leaving them up for too long while you are speaking. Stand Out as a Speaker: Be willing to do shorter talks, such as 30-minute sessions, as many speakers prefer hour-long slots. He notes that having a clever or intriguing title for your presentation is important, and using humor or pop-culture references can help.

Summary: In a unique live recording, Timothy De Block is joined by Sam Chehab from Postman to tackle the intersection of AI and API security. The conversation goes beyond the hype of AI-created malware to focus on a more subtle, yet pervasive threat: "rogue AI agents." The speakers define these as sanctioned AI tools that, when misconfigured or given improper permissions, can cause significant havoc by misbehaving and exposing sensitive data. The episode emphasizes that this risk is not new, but an exacerbation of classic hygiene problems. Key Takeaways Defining "Rogue AI Agents": Sam Chehab defines a "rogue AI agent" as a sanctioned AI tool that misbehaves due to misconfiguration, often exposing data it shouldn't have access to. He likens it to an enterprise search tool in the early 2000s that crawled an intranet and surfaced things it wasn't supposed to. The AI-API Connection: An AI agent is comprised of six components, and the "tool" component is where it interacts with APIs. The speakers note that the AI's APIs are its "arms and legs" and are often where it gets into trouble. The Importance of Security Hygiene: The core of the solution is to "go back to basics" with good hygiene. This includes building APIs with an open API spec, enforcing schemas, and ensuring single-purpose logins for integrations to improve traceability. The Rise of the "Citizen Developer": The conversation highlights a new security vector: non-developers, or "citizen developers," in departments like HR and finance building their own agents using enterprise tools. These individuals often lack security fundamentals, and their workflows are a "ripe area for risk". AI's Role in Development: Sam and Timothy discuss how AI can augment a developer's capabilities, but a human is still needed in the process. The report from Veracode notes that AI-generated code is only secure about 45% of the time, which is about on par with human-written code. The best approach is to use AI to fix specific lines of code in pre-commit, rather than having it write entire applications. Resources & Links Mentioned Postman State of the API Report: This report, which discusses API trends and security, will be released on October 8th. The speakers tease a follow-up episode to dive into its findings. Veracode: The 2025 GenAI Code Security Report was mentioned in the discussion on AI-generated code. GitGuardian: The State of Secrets Sprawl report was referenced as a key resource. Cloudflare: Mentioned as a service for API shield and monitoring API traffic. News Sites: Sam Chehab recommends Security Affairs, The Hacker News, Cybernews, and Information Security Magazine for staying up-to-date.