appsec.fyi

Talks — A Practical Guide

A curated AppSec resource library covering XSS, SQLi, SSRF, IDOR, RCE, XXE, OSINT, and more.

Talks: A Practical Guide

Curated and synthesized by . Last updated 2026-07-01. Synthesized from 85 of 85 curated resources. Browse all 85 Talks resources →

Problem Framing

The application security landscape is in a state of rapid evolution, driven by increasingly complex software architectures, the pervasive integration of AI-generated code, and the distribution of applications across multi-cloud environments [1]. This complexity necessitates a paradigm shift in how security is approached, moving beyond traditional, perimeter-based defenses to embrace a proactive, integrated, and developer-centric strategy. Organizations are grappling with a growing attack surface, sophisticated threats, and the challenge of balancing security with development velocity [1]. Understanding the current threat landscape and the emerging techniques used by attackers is paramount for practitioners aiming to build and maintain secure applications. This requires a continuous learning approach, drawing insights from conferences, research, and practical engagements.

Core Mechanics

The core mechanics of modern application security involve understanding how applications are built, deployed, and attacked. This includes:

Software Supply Chain Security

The foundation of application security often lies in the software supply chain. This encompasses the origin and integrity of all components used in an application, from first-party code to third-party libraries and AI-generated code. The increasing reliance on open-source software and the emergence of AI coding assistants mean that vulnerabilities can be introduced at various points in the development lifecycle [1]. Ensuring the security of the supply chain involves meticulous dependency management, robust build processes, and continuous monitoring for known and unknown vulnerabilities.

Cloud-Native Security

The proliferation of microservices and containerization, particularly within Kubernetes environments, introduces a new set of security challenges. Securing these complex, distributed systems requires a deep understanding of their underlying architecture and the specific threat vectors they present. This includes securing container images, managing admission control policies, and understanding the nuances of network policies and identity management within these environments [2][3].

AI Integration and Security

The integration of AI into the development lifecycle, from code generation to security tooling, presents both opportunities and risks. While AI can accelerate development and automate tasks, it also introduces new attack surfaces. Understanding prompt injection, model backdoors, and the security implications of AI-generated code is becoming increasingly critical [4][5][6][7]. Securing AI workloads requires new security layers and a different approach to threat modeling.

Identity and Access Management (IAM)

In complex, distributed environments, managing identities and access controls becomes a critical security pillar. This extends beyond traditional user authentication to encompass service identities, API access, and fine-grained authorization. Misconfigurations in IAM can lead to significant vulnerabilities, such as unauthorized access to sensitive data [3][8].

Vulnerability Management and Remediation

Effectively identifying, prioritizing, and remediating vulnerabilities is a continuous process. This involves not only scanning for known issues but also understanding the context and exploitability of vulnerabilities to focus remediation efforts. A developer-first mindset, which empowers developers with timely feedback and actionable guidance, is crucial for efficient remediation [1].

Notable Techniques

Several key techniques and threat vectors have emerged as significant concerns for application security practitioners.

Kubernetes Privilege Escalation and Post-Compromise Activities

Attackers are actively exploring techniques to escalate privileges within Kubernetes clusters and then maintain persistence and evade detection. This includes abusing existing services within managed Kubernetes clusters and leveraging middleware components. Post-compromise activities, such as hiding tracks and achieving silent persistence, are also key areas of focus for attackers [2].

Living off the Land in Managed Kubernetes

Attackers can abuse existing services and components within managed Kubernetes environments (GKE, AKS, EKS) to carry out malicious activities without introducing new tools. This involves chaining together existing functionalities to achieve attack objectives [2].

AI-Powered Threats and Exploits

The landscape of AI security is rapidly evolving, with attackers leveraging AI for various malicious purposes. This includes prompt injection attacks, model backdoors, AI-generated phishing campaigns, and AI-powered malware [4]. Research into AI agents for offensive security, aiming for zero false positives, is also a growing area [9].

Supply Chain Attacks on Container Images

Sophisticated techniques are being developed to obfuscate container images, posing challenges for traditional Software Composition Analysis (SCA) tools. Improving SBOM generation algorithms is crucial for enhancing resilience against these obfuscation techniques [3].

Exploiting LLMs and AI Agents

Beyond traditional AI security, specific vulnerabilities are being discovered in LLM applications and AI agents. This includes using simple calendar invites to hijack Gemini agents, leading to actions like deleting calendar events or controlling smart home devices [4]. The secure development and deployment of autonomous AI agents is a major focus, with risks like excessive agency, tool misuse, and prompt injection being actively researched [10].

Vulnerabilities in Peripheral Components

The security of components surrounding core Kubernetes deployments, such as ingress controllers, can be a significant attack vector, as demonstrated by vulnerabilities like #IngressNightmare [3].

Abuse of Cloud Services

Attackers are finding novel ways to abuse cloud services. For example, HazyBeacon campaigns have leveraged stolen IAM credentials to deploy AWS Lambda functions with "AuthType: NONE" for command and control relays [11]. Additionally, vulnerabilities in OIDC issuers allow for hijacking cloud identities by recycling namespaces [11].

Client-Side Attacks on Browser Extensions

Browser extensions, while useful, can present significant attack surfaces. Techniques like DOM injection, closed shadow DOM injection, and clickjacking can be used to compromise extensions. Misconfigurations in web_accessible_resources can also expose internal extension pages [12].

AI Agent Security Masterclass Techniques

A comprehensive approach to attacking and defending autonomous AI systems is emerging. This involves understanding agent architecture, securing Retrieval-Augmented Generation (RAG) pipelines, and threat modeling AI-driven systems. Attack techniques include prompt injection, excessive agency, and tool misuse. Defensive strategies focus on least privilege, sandboxing, and robust prompt controls [10].

Detection & Prevention

Effective detection and prevention strategies require a multi-layered approach, integrating security throughout the software development lifecycle and operational processes.

Proactive Security Measures

A proactive approach to security is essential, moving away from reactive firefighting. This involves integrating security tools early in the development pipeline ("shift left") and embedding security ownership within development teams [1]. Threat modeling, from understanding vulnerabilities to conducting risk prioritization, plays a key role in this proactive stance.

Developer-First Security Tools

Tools that provide immediate feedback and actionable remediation guidance empower developers to address security issues efficiently. Automating routine security tasks frees up developer time to focus on core development [1].

Securing the SDLC

Comprehensive security in the Software Development Lifecycle (SDLC) requires covering all aspects, including first-party code, open-source components, infrastructure-as-code, and containers [1]. This involves practices like building secure container images, utilizing distroless images, and exploring alternative container build solutions when docker build is insufficient [2].

Admission Control and Policy as Code

For Kubernetes environments, implementing admission controllers and policy-as-code solutions is critical for enforcing security policies at scale. While Pod Security Admission (PSA), Open Policy Agent (OPA), and Kyverno each have their challenges, they represent key mechanisms for hardening clusters [2]. Validating Admission Policies offer a more flexible approach compared to PSA [2].

Network Security and Identity

Securing Kubernetes networks requires addressing challenges in Network Policies and the lack of mTLS for internal service communication. Redefining access control, especially for AI agents and services, is an ongoing area of development [3].

AI Governance and Shadow AI Mitigation

As AI adoption accelerates, enterprises face the challenge of managing "shadow AI" – unsanctioned AI tools operating within the environment. Establishing enterprise AI governance and actively monitoring for these tools is crucial, as breaches involving shadow AI can incur significant costs [13].

Secure AI Workload Deployment

Securing AI workloads necessitates new security layers, focusing on the unique risks associated with training and running AI models. This includes protecting against prompt injections and model backdoors [3].

Continuous Monitoring and Threat Intelligence

For Kubernetes, collecting and disseminating threat intelligence through frameworks like eBPF sensors, processed in STIX/TAXII formats, can aid in identifying and responding to threats within the cloud-native ecosystem [2].

Vulnerability Management Prioritization

Accurate risk prioritization, considering exploitability and context, is more effective than simply sending long lists of vulnerabilities to developers. This approach fosters better collaboration between security and development teams [1].

Tooling

A range of tools and frameworks are essential for implementing effective application security practices.

eBPF for Threat Intelligence and Monitoring

eBPF (extended Berkeley Packet Filter) shows promise for collecting threat intelligence in Kubernetes environments. While it has limitations, its ability to act as a sensor for gathering data on system behavior is valuable for security monitoring [2].

Policy-as-Code Tools for Kubernetes

Tools like OPA (Open Policy Agent) and Kyverno are used to implement policy-as-code for Kubernetes clusters, enabling scalable policy deployment and management [2].

Container Image Building Solutions

Solutions for building lightweight and distroless container images, such as those offered by Chainguard, are crucial for reducing the attack surface of containerized applications [2].

AI Security and Red Teaming Frameworks

Frameworks and tools for red teaming AI/ML systems are emerging, focusing on adversarial attacks, security vulnerabilities, privacy breaches, and model manipulation [5]. Tools like Adversarial Robustness Toolbox (ART) and TextAttack are relevant in this space [5].

Vulnerability Management Platforms

Platforms that centralize vulnerability management across cloud, code, and on-premises environments, adding business context for prioritization, are becoming essential [4].

Capture The Flag (CTF) Platforms

CTFs, like Fetch the Flag, provide hands-on environments for practitioners to hone their security skills across various domains, including web exploitation, binary analysis, and more [14][15]. Write-ups from these events offer valuable insights into problem-solving approaches [14][15].

Debugging and Analysis Tools

Tools like Burp Suite are indispensable for web application security testing [16]. For mobile app security, tools like Frida, Objection, and radare2 are crucial for dynamic analysis and instrumentation [17]. For browser extension security, understanding tools for examining manifest.json and debugging content scripts and service workers is key [12].

AI Agent Frameworks and Protocols

Frameworks like LangChain and standards like the Model Context Protocol (MCP) are central to building and securing AI agents. Tools for dynamic, context-aware tool descriptions, like Ragie's Dynamic FastMCP, enhance LLM capabilities [10][6][18].

Recent Developments

The application security field is constantly evolving, with new threats, techniques, and research emerging regularly.

The Rise of AI Security Concerns

AI security has become a dominant theme at major security conferences like Black Hat [4][13]. Discussions revolve around attacking AI models, AI-powered threats, defending AI systems, and using AI for defense. This includes prompt injections, model backdoors, and the security of LLMs and training infrastructure [4]. The rapid adoption of AI is outpacing the understanding and security controls needed to manage it effectively [13].

Agentic AI and Shadow AI Challenges

The complexities of agent-to-agent interactions and the spread of "Shadow AI" are major concerns. Organizations are struggling to govern and secure these emerging AI paradigms [13][7]. Securing autonomous AI agents requires addressing risks like prompt injection, excessive agency, and tool misuse [10].

Kubernetes Security Maturation

While KubeCon continues to focus on security, there's a noticeable shift towards operationalization, secure operations, and enterprise adoption, with fewer purely offensive security talks. This reflects a maturing ecosystem prioritizing stability and scale [3]. However, new challenges in areas like network policies and identity management for AI agents are still being actively researched [3].

Focus on Supply Chain and Component Security

The resilience of Software Composition Analysis (SCA) against container image obfuscation is a continuing area of research, with proposed practical improvements for SBOM generation algorithms [3].

Evolving Threat Models in Cloud Environments

As Kubernetes use cases expand, threat models evolve, driving the need for new types of security controls. This includes securing AI workloads, addressing vulnerabilities in peripheral components, and managing cloud-cluster integration [3].

AI for Offensive Security

Tools and platforms are emerging that leverage AI agents for offensive security, aiming to reduce false positives and automate aspects of penetration testing. XBOW, an autonomous penetration testing platform, demonstrated discovering dozens of vulnerabilities live at Black Hat and DEF CON [9].

AI in the SOC

AI is gaining traction in Security Operations Centers (SOCs) for practical use cases like natural language queries, summarizing investigations, and detection rule translation. However, trust in AI for mission-critical tasks like incident response and containment is still developing [13].

Where to Go Deeper

For practitioners seeking to deepen their understanding of application security, several avenues provide valuable resources and learning opportunities.

Conference Archives and Proceedings

Major security conferences such as KubeCon [2][3], Black Hat [4][19][20][21][22], and DEF CON [23][24][25][26][27][28][29][30] frequently release their talk videos, slides, and white papers. These archives are invaluable for staying abreast of the latest research, techniques, and findings. OWASP also maintains a collection of videos from its global conferences [31][32].

Training and Workshops

Dedicated training sessions are often available at conferences, covering specific domains like AI Agent Security [10], Mobile App Security [17], and various aspects of offensive and defensive security. Platforms like CodeRed offer bundles for AI Mastery in Cybersecurity [33]. Snyk also provides resources like CTF 101 workshops for those new to the field [14].

Online Courses and Educational Resources

GitHub repositories often host comprehensive training curricula, such as those for GenAI security [5] and GenAI essentials [6]. Organizations like OWASP provide extensive resources and project documentation on application security best practices [34][35][36][37][38]. Online learning platforms and YouTube channels dedicated to cybersecurity also offer in-depth content [16][23][39][40][41][42].

Community and Hands-on Practice

Capture The Flag (CTF) competitions, such as Snyk's Fetch the Flag, are excellent for hands-on skill development [14][15]. Participating in or following write-ups from CTFs can provide practical insights into common vulnerabilities and exploitation techniques [14][15].

Research Papers and Technical Blogs

Academic symposia like the IEEE Symposium on Security and Privacy [43] publish cutting-edge research. Following technical blogs from security companies and researchers often provides detailed write-ups and analysis of new vulnerabilities and attack techniques [2][1][44][45][11][46].

Industry-Specific Communities and Initiatives

Organizations like OWASP are central to the application security community, offering numerous global and regional events, projects, and resources focused on various aspects of appsec, including AI security [34][35][7][47][48][36][37][38][49]. Following their activities and participating in their events is highly recommended.

Books and Dedicated Guides

For structured learning, various guides and books on specific topics, such as scaling company security [50] or general cybersecurity roadmaps [51], can provide foundational knowledge and advanced insights.

Sources cited in this guide

  1. Securing next-gen development: Lessons from Trust Bank and TASConnect — snyk.io
  2. Top security talks from KubeCon Europe 2024 — wiz.io
  3. Top security talks from KubeCon Europe 2025 — wiz.io
  4. The insider’s guide to Black Hat 2025 — wiz.io
  5. GitHub - schwartz1375/genai-security-training — github.com
  6. GitHub - schwartz1375/genai-essentials — github.com
  7. OWASP Global AppSec EU 2025 - GenAI Focus — genai.owasp.org
  8. Cybersecurity Conferences 2026-2027 — infosec-conferences.com
  9. XBOW at Black Hat & DEF CON: AI Agents for Offensive Security — xbow.com
  10. AI Agent Security Masterclass: Attacking and Defending Autonomous AI Systems - DEF CON Training — training.defcon.org
  11. [tl;dr sec] #332 - I've Joined OpenAI, fwd:cloudsec, AWS Well Architected Supply Chain Security — tldrsec.com
  12. [HackerNotes Ep.95 & Ep.96] Cookies, Caching & Attacking Chrome Extensions with MatanBer — blog.criticalthinkingpodcast.io
  13. 5 Takeaways from Black Hat x DEF CON 2025 — cybersecuritypulse.net
  14. Sneak Peek into Fetch the Flag CTF 2025 — snyk.io
  15. Fetch the Flag CTF 2026: Official Challenge Write-Ups & Community Highlights — snyk.io
  16. Q&A Session with Snyk & John Hammond: Your Fetch the Flag Questions, Answered — snyk.io
  17. Hacking Android and IOT Apps by Example - DEF CON Training LV 2026 — training.defcon.org
  18. #burp #pentest #ai #hackerassociate #cybersecurity #infosec… | Harshad Shah — linkedin.com
  19. Black Hat USA 2025 Briefings Schedule — blackhat.com
  20. Black Hat USA 2025 - YouTube Playlist — youtube.com
  21. Black Hat USA 2025 Briefings Schedule — blackhat.com
  22. Black Hat Briefings - Wikipedia — en.wikipedia.org
  23. DEF CON 33 Talks - YouTube Playlist — youtube.com
  24. DEF CON 33 Call Index — defcon.org
  25. DEF CON 33 AppSec Village — appsecvillage.com
  26. DEF CON 33 Hacking Conference 2025 — USF — usf.edu
  27. DEF CON 33 (2025) — Security.World — security.world
  28. What to Expect from BSides, Black Hat, and DEF CON 2025 — theregister.com
  29. DEFCON Conference — Official YouTube — youtube.com
  30. DEF CON 33 Archive: Videos, Slides, and White Papers — defcon.org
  31. OWASP Videos — videos.owasp.org
  32. OWASP Videos — videos.owasp.org
  33. AI Mastery for Cybersecurity Professionals — offers.coderedpro.com
  34. BSidesSLC 2026 — sessionize.com
  35. OWASP Global AppSec USA 2025 - CFP — sessionize.com
  36. OWASP Global & Regional Events — owasp.org
  37. OWASP AppSec Days Developer Security Summit — owasp.org
  38. OWASP 25th Anniversary Virtual Conference — sessionize.com
  39. (96) NahamSec - YouTube — youtube.com
  40. Open-Source Intelligence (OSINT) in 5 Hours - Full Course - Learn OSINT! — youtube.com
  41. Fuzzing with Scapy: Introduction to Network Protocol Fuzzing — youtube.com
  42. DEFCONConference — youtube.com
  43. IEEE Symposium on Security and Privacy 2026 — sp2026.ieee-security.org
  44. Silent Breach Lab Writeup (CyberDefenders) — infosecwriteups.com
  45. Connectors CTF 2025 — DFIR Challenges — infosecwriteups.com
  46. AppSec & Cybersecurity Events Calendar 2026: 60+ Conferences — derscanner.com
  47. OWASP Global AppSec EU 2025 (Barcelona) — owasp.glueup.com
  48. OWASP Global AppSec USA 2025 (Washington DC) — genai.owasp.org
  49. The Elephant in AppSec Conference — theelephantinappsec.com
  50. Gibler - An Opinionated Guide to Scaling Your Company's Security - Google S — docs.google.com
  51. Cybersecurity Slides Collection — github.com
📚 This guide is synthesized from the full text of resources curated in the Talks library, and refreshed as new material is added.