Who governs the agents? MCP, multi-agent architectures and control risks
20 October 2026 | 12.30h to 13.00h
This session will examine the security risks associated with multi-agent systems based on MCP and Agent-to-Agent (A2A) communications, highlighting threats such as malicious MCP servers and the lack of role- and permission-based controls. It will explore how these weaknesses can lead to misuse, unauthorized execution, and chained attacks across multiple agents.
The presentation will take a practical approach, using real-world architectures and scenarios to illustrate how risks can propagate through multi-agent systems. It will also cover mitigation strategies, governance frameworks, and regulatory compliance aligned with the AI Act. Finally, it will present a holistic perspective that combines technical security and legal accountability to support the safe deployment of these technologies.
Presenter:
MIREYA SALUDES
Head of business development and ecosystem, CIDAI
Professional with more than 20 years of experience leading multidisciplinary teams and driving innovation and digital transformation processes in public and business environments. Specialized in developing growth strategies, defining new value propositions and creating key alliances for business development. With a solid track record in generating and energizing ecosystems, she has led projects with high territorial and sectoral impact, combining strategic vision and operational execution.
Speakers:
LAURA APARICIO ALBA
Senior Legal Consultant, Conversia
Laura Aparicio Alba holds a Law degree from the University of Girona and is a certified Data Protection Officer (DPO) by the Spanish Data Protection Agency. She has over nine years of experience in compliance consulting.
She currently leads multidisciplinary teams and participates in the development of legaltech solutions aimed at automating, improving the efficiency and scalability of legal services. In the field of artificial intelligence, she specializes in the application of governance, risk management, and compliance (GRC) frameworks, promoting responsible, ethical, and secure use of AI within organizations.
VÍCTOR GARCÍA DOMÍNGUEZ
Head of Artificial Intelligence, Conversia
Víctor García Domínguez holds a degree in Nanoscience and Nanotechnology from the UAB and completed his thesis at Florida International University. He also holds a Master’s degree in Visual Tools to Empower Citizens from the University of Girona (UdG) and a Master’s degree in Data Science from the UOC.
He currently leads the Artificial Intelligence team at Conversia, where he has been driving the deployment of generative AI solutions for the past three years. He has also delivered comprehensive corporate training programmes that integrate AI expertise across organisations. In addition, he has actively participated in numerous conferences and technology projects, contributing to programme design, event organisation, and delivering talks on applied artificial intelligence.