DEF CON 34: Jonathan Brossard presents his work on automated vulnerability discovery

MOABI 12 August 2026

DEF CON holds a special place in the world of cybersecurity. For over thirty years, the conference has brought together security researchers, hackers, engineers and offensive security specialists in Las Vegas to present their discoveries, tools and work to a particularly discerning international community. Being selected to present one’s research there is therefore a significant honour. This year, Jonathan “endrazine” Brossard, CTO of MOABI, was once again a speaker at DEF CON, delivering a talk on his latest work: Witchcraft Solver: Automated 0-day Discovery in Stripped Binaries. Start with a binary and automatically scan for vulnerabilities The problem that Witchcraft Solver tackles is simple to describe, but considerably more difficult to solve. Let’s take a compiled piece of software for which we do not have the source code. The symbols have been stripped and no specific environment has been set up to facilitate its analysis. How can we automatically search for vulnerabilities in this software and proceed until we obtain a proof of concept? This is precisely the aim of the work presented by Jonathan at DEF CON. Witchcraft Solver offers an automated pipeline combining several advanced cybersecurity research techniques. The binary is first transformed to enable its analysis in an intermediate LLVM representation. An initial analysis then reduces the number of relevant targets before several formal verification engines are run in parallel. The results obtained are then fed into directed fuzzing and concolic execution techniques to compare potential vulnerabilities with the programme’s actual behaviour and move towards concrete proofs of concept. The challenge is therefore not simply to detect an anomaly. It is to gradually narrow the gap between an unknown binary and the demonstration that vulnerable behaviour actually exists. Comparing research with real-world software Another important aspect of the work presented concerns its application to environments representative of real-world industrial settings. According to the results presented by Jonathan, the multi-architecture approach studied was tested against a corpus of 39,364 production ELF binaries and supports various architectures, including ARM64, ARMv7, RISC-V and s390x. The pipeline was also tested against a known vulnerability in libjpeg-turbo, CVE-2023-2804. In the experiment presented, the use of SymQEMU enabled the first crash to be achieved within 25 minutes from a stripped binary, without access to the source code. This work thus explores a particularly important question for security research: to what extent can we automate tasks that have historically been the preserve of experts with advanced skills in reverse engineering and vulnerability analysis ? An issue that goes far beyond academic research This issue also has a clear industrial dimension. Modern digital products are made up of numerous components. Proprietary software, open-source libraries, firmware, third-party components and vendor-specific developments coexist within systems whose complexity is constantly increasing. And the company that uses or integrates these components does not necessarily have access to their source code. Yet it must be able to assess their security. This issue becomes particularly important in industrial, embedded or mission-critical environments, where certain equipment may remain in production for many years and incorporate components from multiple suppliers. Analysing the software as it is actually delivered therefore becomes essential. This is also what makes research into binary analysis particularly interesting: it allows researchers to work directly on the final artefact, without necessarily relying on the development environment that produced it. A particularly significant return to DEF CON This talk forms part of a long-standing relationship between Jonathan Brossard and DEF CON. Jonathan previously presented his research on BitLocker and TrueCrypt in 2008, Rakshasa in 2012, and his work on the Witchcraft Compiler Collection in 2016. His return in 2026 with Witchcraft Solver thus continues nearly twenty years of contributions and work in the fields of offensive security, reverse engineering and vulnerability research. His career has also seen him present his research at several major international conferences in both the industry and academic sectors, including Black Hat, CCC, HITB, IEEE, USENIX and ACM. He is currently completing a PhD on reverse engineering at CNAM, where he has also taught for several years. For MOABI, seeing our CTO return to present his research at DEF CON is naturally a source of great pride. But beyond individual recognition, this presentation above all illustrates a conviction: developing advanced cybersecurity technologies requires maintaining a close link between research, experimentation and real-world industrial challenges. Gaining a deep understanding of software, exploring new methods of analysis, automating what once seemed difficult to automate, and sharing the results with the international community are all part of this approach. Research must not remain confined to laboratories or conferences.It really comes into its own when it gradually enables us to tackle specific security issues. Having attended in 2008, 2012 and 2016, Jonathan Brossard was back at DEF CON in 2026. This time, he posed a particularly ambitious question: to what extent can we automate the search for vulnerabilities when all we have is the binary ?

VivaTech 2026: Innovation Can No Longer Move Forward Without Cybersecurity

MOABI 16 June 2026

The 2026 edition of VivaTech comes at a pivotal moment for the European digital ecosystem. Software has never been more central to innovation. Artificial intelligence, cloud computing, connected devices, embedded systems, Industry 4.0 and smart mobility all rely on thousands of interconnected software components, many of them developed by third parties and integrated at increasing speed. This acceleration is creating tremendous opportunities. It is also creating a significant challenge: managing risk. Over the past few years, cyberattacks have shown that organisations are no longer targeted solely through their own systems. Increasingly, attackers are exploiting suppliers, software dependencies and trusted components embedded deep within digital supply chains. Recent supply chain attacks have demonstrated how a single compromise can propagate across entire ecosystems and affect thousands of organisations simultaneously. The growing complexity of modern software environments is making visibility and control more difficult than ever. At the same time, European regulations continue to raise the bar. The Cyber Resilience Act, NIS2 and other regulatory frameworks are introducing new expectations around product security, vulnerability management and software supply chain governance. Cybersecurity is no longer viewed as a purely technical matter. It is becoming a business, compliance and resilience requirement. This evolution raises important questions. How can organisations maintain visibility over increasingly complex software systems? How can they manage risks introduced by third-party components and suppliers? How can they keep pace with the constant flow of newly disclosed vulnerabilities? How can innovation remain agile while maintaining a strong security posture? These are some of the questions that will undoubtedly shape discussions throughout VivaTech 2026. This year’s event is expected to place a strong emphasis on artificial intelligence, digital trust, sovereignty, cybersecurity and resilient digital infrastructure, reflecting the priorities currently driving technological transformation across Europe. As a premium partner of the event, Orange is highlighting its vision of a trusted digital future, bringing together startups, industrial players and technology leaders to explore the next generation of secure and innovative digital services. More than sixty startups from the Orange innovation ecosystem will be showcased throughout the event. In a world increasingly dependent on software, one idea is becoming clear: Trust alone is no longer enough. Organisations must be able to understand, assess and verify the technologies on which they rely. The next decade of innovation will not be defined solely by performance or new capabilities. It will also be defined by trust, resilience and the ability to build secure digital foundations for the future.

DeepTech Forum Sophia Antipolis #2: A Look Back at a Day of Ecosystem Building, Embracing Failure, and Shaping the Future

MOABI 3 June 2026

On June 3, we were invited to participate in the second edition of the DeepTech Forum Sophia Antipolis, organized by students in the Specialized Master’s in Deeptech Entrepreneurship and Innovation program at Mines Paris-PSL, in collaboration with Dynergie. It was a day dedicated to the ecosystem, open dialogue, and collective reflection on the future of disruptive innovation in the region. With a booth on the Pierre Laffitte Campus, we took advantage of the day to meet the Côte d’Azur deeptech community, share our approach to product cybersecurity, and talk with those who drive the ecosystem on a daily basis: entrepreneurs, researchers, investors, support organizations, students, and public sector representatives. At the booth, Nicolas conducted a series of demos of the MOABI platform, discussed our client use cases, and engaged in broader conversations about the role of a deep tech entrepreneur. A direct and rigorous approach to disruptive innovation The tone of the event, set from the outset by its patron Emilie Royère, director of Eurobiomed, did not seek to sugarcoat reality. This second edition chose to address disruptive innovation head-on, through candid and at times uncomfortable discussions. The most memorable panel discussion of the afternoon, titled “Failing in Deep Tech: Luxury, Choice, or Waste?”, clearly illustrated this approach. Failure in deep tech often remains a taboo subject, even though it is an integral part of the journey. Being able to discuss it publicly, with entrepreneurs who have weathered the storm and investors willing to speak candidly, is exactly the kind of conversation the ecosystem must be able to have if it is to mature. Later, the joint presentation by Euronext and Enogia traced the concrete path of a deep-tech company, from fundraising to its initial public offering. This is a rare journey in the French deep-tech innovation landscape and a valuable case study for the founders in attendance who are looking toward the long term. Sophia Antipolis 2035: Moving Beyond Denial The afternoon concluded with regional foresight workshops focused on the future of Sophia Antipolis and the French Riviera by 2035. We participated in the first workshop, whose topic left no room for complacency: Sophia has missed the deep tech curve. Why, and how can this be corrected? A direct, almost provocative statement that had the merit of sparking a frank discussion among students from Mines, active entrepreneurs, and local stakeholders in the technology park. Several key themes emerged from the discussions: the historical difficulty in transforming regional scientific excellence into industrial projects, the scarcity of the patient capital required for deep tech, competition from other better-structured French and European ecosystems, and the need to build stronger bridges between laboratories, specialized training programs, and established manufacturers. The exercise was not intended to produce an action plan. Its value lay elsewhere: collectively acknowledging the reality of the situation, without trying to sidestep it, is likely the prerequisite for any serious course correction. What we took away from the day Beyond the conferences and workshops, it was the quality of the informal exchanges that left a lasting impression on us. Whether at the booth, over coffee, or during the closing cocktail reception, we had lengthy conversations with founders who shared our concerns about growth, funding, recruitment, and the challenges of long-term tech entrepreneurship. A true convergence of perspectives on what it means to build a deeptech company in France today. We left the Pierre Laffitte campus having forged connections, identified follow-up projects, and with a strengthened conviction that this type of event, on a human scale and resolutely ecosystem-oriented, is essential to the maturation of regional deeptechs. A big thank you to the students of the Specialized Master’s in Deeptech Entrepreneurship and Innovation at Mines Paris-PSL, who designed and organized this event with a sense of community and a level of seriousness rarely seen at this stage of their education. Thanks also to Dynergie for its support, to Emilie Royère for sponsoring this edition, and to all the speakers and participants who made this day so enriching. See you next year, we look forward to it.