Speakers
Adrian Novoa – RedTeam 2026: Initial Access in Modern Environments
b0t – Watch This! Writing malicious Bluray Discs
cesi0 – What Your Car Knows: Digital Forensics and Modern Infotainment Systems
Eliad Kimhy – The History of Malware: 40 Years of Viruses, Worms, and Other Strange Creatures
Jack Hooker & Erwin Karincic (Dollarhyde) – Hacking Drones for Fun and Knowledge
jk – Hacking Doodleburg – Reverse Engineering Electronic Children’s Toy Firmware
Joe Rozner – Vibes all the way Down
Larry Suto – Hunting Rogue Autonomous Agents: A Field Guide to AI Threats the SOC Can’t See
Mehmet Sencan – Pulverize Chips with a Squib in the Heatsink
Mohsen Ahmadi – SPECTRA: Semantic Pattern Extraction for Scalable Malware Detection and Signature Automation
Priyank Nigam – Trusted Senders and Untrusted Outcomes: Modern HTML Injection Exploitation
Tobias Mueller – Busting Black Basta: Exploting Ransomware
RedTeam 2026: Initial Access in Modern Environments
Adrian Novoa
In today’s modern enterprise environments, as professional Red Team and Offensive Security engineers, we can no longer rely on the simplistic tactics of the past where delivering a single malicious file was sufficient to establish an initial foothold.
The threat landscape has evolved dramatically, and so must our approach. Our current focus centers around creative social engineering campaigns, the use of obscure and unconventional file attachments to evade detection, and the exploitation of hybrid environments where a single compromised token can yield significant unauthorized access across an organization’s infrastructure.
In this presentation, we will provide an overview of several of these techniques, introduce new tooling developed to support modern offensive operations, and take a deeper dive into Microsoft Azure examining precisely what level of access a compromised access token can grant an attacker, and what that means for the organizations we are tasked with protecting.
Helpdesk roots, offensive security mindset. Certified in CRTO and Cyberwarfare Labs, currently working at Init1Security. Music aficionado, strength trainer, and dedicated father.
(^Top)
Watch This! Writing malicious Bluray Discs
b0t
The Bluray disc format is widely used, even in today’s world of streaming. However, Bluray discs and players have properties that can make them useful for more than playing movies. In this talk, we’ll go from a cute idea to code running on every major Bluray device including game consoles. We’ll talk about how to install your own code on a bluray player and how you might leverage such a capability for fun and profit.
b0t is a malware and vulnerability researcher with over 10 years of experience. He creates and publishes VX-Underground’s Black Mass malware research publication and has done malware and security research work professionally at SpaceX and AWS. b0t has previously taught classes on rootkit design, malware design and peer to peer networks.
(^Top)
What Your Car Knows: Digital Forensics and Modern Infotainment Systems
cesi0
Have you ever thought about what your car knows about you? What about your devices, contacts, and even the messages you’ve sent and received? Imagine what someone with access to this system could do. Do you trust that valet? Should you? In this talk, we discuss the myriad of information that modern infotainment systems collect and even perform a deep dive into the recovery of such data.
Threat Hunter. Incident Responder. Detection Engineer. All-around degen. As a 15 year veteran of the security industry, cesi0 has done forensics, malware reverse-engineering, threat analysis, red teaming, and responded to incidents ranging from banal to felonious. Returning to LayerOne for his 12th year, cesi0 brings extensive experience in all things threat.
(^Top)
The History of Malware: 40 Years of Viruses, Worms, and Other Strange Creatures
Eliad Kimhy
In 1984, a columnist for Scientific American wrote about a story he’d heard: two programs, Creeper and Reaper, hunting each other across a corporate network. He admitted the story had obvious holes in it, but he believed it anyway, “”perhaps because I wanted to.”” A year later, readers had sent him so many real examples of viruses that he described the possibilities as “”so horrifying that I hesitate to set them down at all.”” In the years to follow, malware will make the remarkable shift from myth to reality, and it would not stop there.
The history of malware is weirder, funnier, and more instructive than most people realize. A ninth-grader wrote one of the first Apple II viruses as a prank because his friends had wised up to his tricks and refused to take any of his floppy disks. A macro virus from the late 90s would hijack your Word document, type a message on screen, and then play a generated audio file of someone laughing at you. A botnet author in 2005 was releasing variants so fast he started leaving angry messages for antivirus researchers embedded in the code, furious that they kept misspelling his malware’s name.
You might have heard some of these stories, but there are far more that haveremained untold. This talk is a research-driven walkthrough of over 40 years of malware history, built from primary sources and forgotten security publications: decades of Virus Bulletin magazine archives, Byte Magazine articles from the late 80s, early security research papers, and the actual words of the people who wrote these viruses. We’ll trace the full arc from Creeper in 1971, through boot sector viruses spreading at the speed of sneaker shoes, the macro virus epidemic that overtook all other malware within three years of Office ‘95 releasing, the convergence of worms and remote access trojans that created the first botnets, banking trojans that learned to drop ransomware, and the multistage attack chains we deal with today.
But underneath the strange stories, there’s a consistent pattern. Each major technology shift (personal computers, the internet, file sharing, networked enterprise, and more recently, large language models) triggers the same cycle: experimentation, creative mischief, and then real weaponization. Many of the techniques we think of as modern had all existed at one point or another, in one form or another. The basic principles of social engineering have barely changed since 1989. Macro viruses, one of the oldest attack vectors, are still in active use. And for our part, we seem to consistently forget the lessons of the past, whenever a new-fangled technique comes on the scene.
The talk closes by turning this historical lens toward the present moment. If every major technology shift has followed this pattern, and generative AI is the defining technology of this decade, where are we in this cycle? And what does 40 years of malware history suggest about what comes next?
Eliad Kimhy is a Senior Security Researcher at Acronis, where he conducts research into emerging threats and cybercrime, and shares insights through conference talks and published reports. Eliad has worked with security teams for close to a decade, helping build and lead the development of threat intelligence production, and the publication of research-based content for technical and general audiences. He has spoken at conferences such as VirusBulletin, CARO, Insomnihack, Thotcon, BsidesSF, BsidesLV, and IT-SA. He is the co-creator and producer of the Webby Honoree podcast Malicious Life, which explores the untold stories and cultural history of hacking.
(^Top)
Hacking Drones for Fun and Knowledge
Jack Hooker & Erwin Karincic (Dollarhyde)
The drone market is growing fast, but UAS security hasn’t kept pace. Unlike most IoT devices, drones operate in physical space, which means vulnerabilities don’t just leak data, they create kinetic risk. This talk digs into the current state of UAS security testing, examining the full attack surface of modern commercial drone ecosystems across software, RF, and protocol layers, from the aircraft itself to the broader command and control infrastructure.
Taking a full-stack approach, we present research on practical offensive techniques, we cover RF-layer exploitation, attacks springboarded from end user devices, and abuse of broadcast protocols. We discuss interesting discoveries in UAS AI capabilities and Remote ID weaknesses. Attendees will leave with a clearer picture of how modern drone ecosystems actually break and how to start testing them. Ultimately, this research serves to improve UAS security and inform future standards and regulations so that these classes of vulnerabilities don’t find their way into the next generation drones.
Jack is a security researcher, specializing in IoT, cloud security, and unmanned aerial systems (UAS). His past projects have included research into UAS Detection, UDP amplification attacks and cell site simulator detection. When not aggressively probing client networks, Jack spends his time building, flying, and crashing UAS. He also brings extensive experience in hardware design, embedded systems, and rapid prototyping to the team. Jack holds numerous certifications in these disciplines, including OSCP, OSWP, and AWS CSA.
Erwin is an experienced security researcher specializing in both hardware and software reverse engineering, binary analysis, and exploit development across a range of processor architectures. He has notable experience in implementing complex Radio Frequency (RF) waveforms using Software Defined Radios (SDRs) for cybersecurity applications, complemented by his proficiency in designing, simulating, and fabricating antennas tailored for such applications. His past work includes extensive TCP/IP networking experience, designing worldwide secure communication systems. Erwin holds a number of prestigious certifications, including OSCP, OSCE, OSWE, OSEE, and CCIE Enterprise Infrastructure.
(^Top)
Hacking Doodleburg – Reverse Engineering Electronic Children’s Toy Firmware
jk
A couple of years ago, at LayerOne, a handful of bored hackers walked to Target and browsed the toy aisle. On a whim, they bought an electronic kids’ toy called Mr. Pencil’s Scribble, Write, and Read. After using a ripped up ethernet cable and a borrowed Arduino to build a SPI firmware dumper, the real journey began. Come along for a story about hardware reverse engineering, 65C02 firmware, weird bank-switched flash ROM, custom graphics and animation formats, custom audio codecs, fun with hydrofluoric acid, and several hundred strings of profanity in this children’s toy. After 2 years, will our heroes finally get arbitrary code execution on Mr. Pencil?
Jon King (jk) is a security researcher with a fondness for firmware reversing and emulation, ropping for the joy of it, and using JB Weld and duct tape as a replacement for proper tools. He also enjoys developing defeats for high-security locks and tamper-evident devices. You can find him in the lobby at 3AM tricking the LayerOne demoboard into doing things it’s not supposed to do.
(^Top)
Joe Rozner
Vibes all the Way Down covers observations from using LLMs to do reverse engineering and vulnerability research against two targets: ionCube and the Flipper Zero. ionCube is a PHP source protection tool that compiles scripts to encrypted bytecode and ships a loader as a PHP extension. I first encountered it in 2022 while bug hunting a corporate spyware vendor’s backend, where I paid a sketchy service to decompile the files and used the recovered source to identify an exploitable pre-auth chain. In early 2026 I revisited the exploit chain and set out to build my own ionCube decompiler with AI assistance, then applied the same workflow to the Flipper Zero to identify an exploitable remote code execution delivered over NFC. The presentation walks through how I approached each project, the results, lessons learned, 0day releases, and specifics of the tooling and workflow.
Joe does offensive security and some other stuff. At one point he was a big enough deal to have a Google search AI summary.
(^Top)
Hunting Rogue Autonomous Agents: A Field Guide to AI Threats the SOC Can’t See
Larry Suto
A state-sponsored group recently used an AI coding tool to autonomously conduct 80-90% of a cyber espionage operation targeting 30 organizations. The lab behind the tool then disclosed that its next unreleased model is “currently far ahead of any other AI model in cyber capabilities.” Exploitation timelines have collapsed from 771 days to four hours, and over $700 million in venture capital is now funding autonomous offensive security platforms. Yet AI agent testing remains largely manual, single-shot, and without a shared taxonomy for what to test for.
This talk changes that. It introduces SASTER (Systematic Agent Security Threat Evaluation and Rating) a 24-pattern taxonomy for classifying rogue agent behaviors across three tiers: overt deception, “”correctly wrong”” behaviors, and tool trust exploitation. SASTER patterns map to the OWASP AIVSS agentic AI risk framework, enabling practitioners to score findings using industry-standard severity metrics. Three live demonstrations show the patterns in action: a multi-turn conversation attack that defeats defenses no single prompt can breach, a systematic probe campaign revealing where defenses hold and where they collapse, and real-time agent-to-tool traffic interception catching rogue behaviors as they happen. The talk then walks through how to build a practical SASTER test harness probe library, scoring function, multi-turn orchestrator, traffic interceptor, and scorecard output that attendees can implement against their own deployments in a weekend.
Larry Suto is a security researcher currently building autonomous red teaming tools for generative AI systems. His current research covers multi-turn attack methodology, rogue agent detection taxonomy (SASTER – 24 patterns across three threat tiers), and AI defense assessment methodology
(^Top)
Pulverize Chips with a Squib in the Heatsink
Mehmet Sencan
We’ll cover how to destroy chips instantly, completely, legally and safely, and give pointers on what did or did not work for us. As TEEs in high-performance computing hardware become increasingly powerful and valuable targets for espionage and sabotage, protecting the intellectual property, cryptographic keys, and sensitive data they contain is of paramount importance. This talk argues physical destruction provides stronger guarantees than other methods, such as zeroization, but unlike custom-engineered destructive solutions such as PyroMEMS nanothermite, our approach leverages existing industrial components with proven reliability. This significantly reduces the complexity and cost of the implementation. We demonstrate that a common squib, like those used to simulate bullet hits in movies or open flaps in aircraft, when appropriately positioned within a modified GPU heatsink, can provide effective physical destruction of the computing hardware. The proposed solution offers a balance of effectiveness, cost, reliability, and implementation simplicity that makes it suitable for immediate deployment in secure computing environments.
Mehmet is taking a hardware backstop approach to security and governance of AI compute. Since finishing his BS at Caltech in Applied Physics, he has been pushing chip and manufacturing technology capabilities for over a decade, previously as a full-stack hardware developer, running biosensor manufacturing processes all the way from sensor design to medical device implantation (while ensuring functionality, cost-efficacy, and manufacturability).
(^Top)
SPECTRA: Semantic Pattern Extraction for Scalable Malware Detection and Signature Automation
Mohsen Ahmadi
Modern malware detection relies heavily on behavioral signatures, yet the process of developing these signatures remains largely manual, slow, and fragile in the face of rapidly evolving variants. Minor syntactic changes such as recompilation, control-flow obfuscation, or API indirection are often sufficient to evade traditional similarity techniques based on hashes, opcode frequency, or structural features. This creates persistent detection gaps and delays in response.
This talk presents SPECTRA, a system designed to transform malware analysis into a scalable, semantic-driven pipeline for automated detection engineering. At its core, SPECTRA introduces the concept of a binary semantic genome, a compact representation of invariant program behavior extracted through VEX IR normalization, static dataflow analysis, and adversarially robust feature weighting.
We demonstrate how SPECTRA leverages a Rust-based analysis engine (Binlex) to lift binaries into VEX IR, enabling architecture-agnostic semantic extraction at scale. From this representation, the system constructs multi-view fingerprints incorporating semantic transformations, dataflow relationships, API interactions, and string embeddings. These fingerprints enable large-scale similarity search across corpora, identifying previously undetected variants and quantifying detection coverage gaps. We evaluate SPECTRA against existing similarity approaches, including context-aware fuzzy hashing techniques such as Ethos, TLSH, and other structural similarity hashes commonly used in platforms like VirusTotal. While these methods improve over traditional hashing, they remain sensitive to semantic-preserving transformations. In contrast, SPECTRA’s semantic genome provides a more robust and stable representation, enabling reliable detection of related variants even under obfuscation and structural mutation.
To close the loop, SPECTRA integrates constrained large language models to automatically generate candidate behavioral signatures and corresponding validation tests, ensuring compatibility with production detection engines. We will discuss system architecture, similarity metrics, and evaluation strategies, and show how combining semantic normalization, scalable analysis, and constrained AI enables a shift from reactive signature creation to proactive, data-driven detection engineering.”
Mohsen Ahmadi is a security researcher with the Cisco Talos Threat Research team specializing in malware analysis, reverse engineering, and scalable detection engineering. He previously worked on Apple’s Platform Architecture Security team focusing on M1 chipset architecture security. He has presented at the Google Security Summit and serves as a reviewer for security conferences, including the Artifact Evaluation Committee for IEEE S&P. Mohsen studied Computer Science at Arizona State University and has participated in DEF CON as part of the Shellphish CTF team.
(^Top)
Trusted Senders and Untrusted Outcomes: Modern HTML Injection Exploitation
Priyank Nigam
This talk walks through new offensive techniques for initial foothold avoiding malicious attachments (which have high rate of detection) and focus on TTPs for weaponizing trusted cloud workflows, including web API manipulation targeting email clients where JavaScript execution is disabled, creative HTML injection paths, filter bypass strategies, and input-length constraint abuse. I’ll also cover the dead ends where exploits that looked promising but later collapsed under real-world conditions. This was inspired from a recent red team operation, after burning through the usual initial-access vectors where I pivoted toward a less explored attack surface: cross-tenant abuse of Microsoft online services that can send email on behalf of a highly privileged service principal. Because the payloads were delivered through a legitimate Microsoft 365 service, inbox placement was effectively guaranteed, and user-click probability is thus high. We’ll dissect how desktop email clients differ from their web-based counterparts, and how those differences can be leveraged to shape exploit delivery and execution. Although the specific vulnerability was responsibly disclosed and quickly patched, the underlying patterns are far from unique and similar weaknesses likely exist across other enterprise applications. We’ll close with practical guidance for defenders on building hardened email-templating and notification pipelines, ensuring that your own web services can’t be hijacked by threat actors to gain initial access or escalate their foothold.
As an senior security engineer, Priyank’s primary areas of focus is conducting security exercises that emulate real-world threats impacting billions of users. He is well-known for his expertise in identifying high-impact vulnerabilities and has shared his research openly through various industry conferences. His forte is full-stack security assessments via threat modelling, penetration testing and secure source code reviews. In the past, he has advised Fortune 500 brands and startups and does mobile and IoT related research in his spare time. As a new parent, he is now (re)learning hacking from his toddler(s) who defeat all the “restrictions” to do whatever the hell they want to.
(^Top)
Busting Black Basta: Exploting Ransomware
Tobias Mueller
With a turnover of multiple million USD per year, the Russian ransomware crew Black Basta was one of the biggest threat actors attacking over 500 organisations world wide. But even though they certainly would have had the resources to do proper QA, their software had bugs which ultimately allowed for the recovery of encrypted files without paying the ransom. This talk describes the threat actor, their flaw, and the recovery.
We start with a technical deep-dive into the encryption flaws of the Black Basta ransomware. These flaws helped to recover files encrypted by of the most prolific ransomware crews without having to pay the ransom.
We further explore vulnerabilities in the treat actors infrastructure and discuss a novel way to get shell via SSH through accounts that were thought to be protected trough authorized_keys’ “command” stanza. This command execution technique enabled a unqiue view into the threat actor’s tactics and procedures as well as their infrastructure.
We finish with recommendations to tactics and procedures commonly employed by ransomware threat actors to provide some idea of how to actually reduce the attack surface.
Tobias Mueller is a cybersecurity expert with over a decade of experience spanning academia, applied research, and consulting. He holds a PhD in cryptography, is a CISSP, CISM, and OSCP as well as a certified ISO 27001 auditor. Tobias supports clients in the financial and insurance sectors with red teaming, incident response, SOC design, compliance, and third-party risk management.
(^Top)
