Dreadnode
Back to Newsroom
Event

Dreadnode at BlackHat, DEFCON, & AI Security Forum: Own Your Autonomous Security Stack

By Dreadnode Crew

New ScopeJudge research, a self-improving red-vs-blue Briefing, and platform updates — here’s where to catch Dreadnode across Black Hat, the Las Vegas AI Security Forum, and DEF CON 34.

WHO: Dreadnode — the AI infrastructure platform for full visibility and control over your cyber capabilities, from AI red teaming to web security to network operations. Research and leadership team will be on-site, available for briefings.

WHAT: A Black Hat Briefing, released research, a platform update, and talks across three venues — all built on one idea: security teams should own their autonomous capabilities, not rent black-box agents. Build it, prove it, ship it, own it — and keep it on the rails.

WHEN: August 4–9, 2026

WHERE: Mandalay Bay (Black Hat, Aug 4–6) · The Westin (Las Vegas AI Security Forum, Aug 6) · Las Vegas Convention Center (DEF CON 34, Aug 7–9)

On-site highlights

  • Platform updates. Demo our cyber capabilities, with a focus on multi-agent, horizontally scaled workflows for use cases like vulnerability research, and operator-driven, human-in-the-loop pentesting and bug hunting. Get a first look at the latest release featuring persistent memory and continual learning, structured agent output (pre-defined Assets and Findings, or your own custom types), and Claude Code support — with support for other harnesses coming soon.
  • AI red teaming. Demo the latest updates to our AI red teaming capability, which focus on assessing multimodal models for security and safety vulnerabilities (we added 100+ transforms dedicated to this), agentic systems for security vulnerabilities, and traditional ML models for evasion, extraction, and membership inference attacks.
  • ScopeJudge research. A benchmark of 4,897 tool calls, labeled call-by-call by professional penetration testers, testing whether an inexpensive, trusted judge can stop an autonomous agent before it acts out of scope. Headline finding: a fixed policy can’t do it — scope lives in the operator’s request. Dataset will be released to the community.
  • Black Hat AI Summit (Tue Aug 4). Tuesday, August 4, 2026 · Mandalay Bay, Las Vegas, NV · Oceanside A + D, Level 2. Co-Founder Will Pearce joins the panel “From AI Assistance to AI Autonomy: Security Lessons Learned” (Tue Aug 4, 4:35–5:15 pm, Oceanside A, Level 2), alongside Nathan Hamiel (Senior Director of Research, Kudelski Security), Victoria Westeroff (Principal Director, AI Red Team Ops, Microsoft), and Jess Burn (Principal Analyst, Forrester Research). AI has shifted from decision support to operational actor, forcing security teams to redefine ownership, accountability, and trust faster than their defenses are evolving. Agentic systems, token economics, and AI-on-AI attacks are introducing new attack surfaces that traditional SOC, governance, and risk models are not yet equipped to manage — the next phase of AI security demands tighter human oversight, clearer operating models, and a decisive move from experimentation to measurable, adversary-tested resilience. Session details. · AI Summit info.
  • Black Hat Briefing: “Catch Me If You Can: AI Investigators Hunting Autonomous Attackers as a Benchmark” (Thu Aug 6, 12:00 pm, Oceanside B, Level 2). Coordinated red and blue AI agents on live, production-like enterprise infrastructure: the autonomous attacker takes full domain dominance across a 3-forest Active Directory environment in under 20 minutes, and its traces become ground truth for grading the defenders. Speakers: Martin Wendiggensen and Jayson Grace. Full talk details.
  • Las Vegas AI Security Forum (Thu Aug 6, The Westin). Dreadnode will be kicking off the AI Security Forum with local coffee shop Mothership — we’ll be serving premium espresso drinks from 8–10 am PT during registration. Plus, CTO and Co-Founder Nick Landers will be leading a workshop titled “Hands-On Offensive AI from Primitives to Capability” (time and location TBD). In this workshop you’ll build, evaluate, and iterate on an offensive cyber capability end-to-end. In under 90 minutes you’ll run it against live targets, read its trajectory step by step, modify its tooling and scaffolding to fit your own operational requirements, and wire up a self-improvement loop that grades runs and feeds the results back in. Request an invite.
  • DEF CON 34 (Aug 7–9). Catch Ads Dawson across three villages: “Exfil Everything: A Year of Stealing Data from AI Agents” with Mike Takahashi (Bug Bounty Village), “Pentesting AI Is Not a Vibe Check” (OWASP Village), and a keynote panel (Red Team Village, Fri). At AI Village, Vincent Abruzzo will be presenting the AgentLens agent failure modes research paper co-authored by Google DeepMind’s Neel Nanda.
  • Offensive AI Con Black Hat Party (Wed Aug 5). Join us at the official Offensive AI Con Black Hat Party on Wednesday, August 5, 7 pm–midnight at the BrewDog rooftop. Request an invite.

Spokespeople available for briefings: Brad Palm, Nick Landers, Will Pearce, and the research teams behind the Briefing, ScopeJudge, and the various DEF CON sessions.

To schedule a briefing or on-site meeting: Tori Norris, tori@dreadnode.io or contact us.

About Dreadnode

Own your autonomous security stack with Dreadnode. Founded by offensive security and AI red team leads from NVIDIA, Microsoft, and NetSPI, the Dreadnode Platform enables teams to build, evaluate, and operationalize models and agents you can trust in production. Whether you’re finding vulnerabilities before adversaries do, accelerating network operations, or AI red teaming at machine speed, Dreadnode provides the infrastructure to make it happen. We envision a world where security costs less and does more. Dreadnode’s efforts to make AI-native security a reality are supported by Decibel, In-Q-Tel (IQT), NFC, Sands Capital, Indie VC, and Aviso Ventures.