Security, Trust & ComplianceLunch Talk40min
Who to blame? The AI, The Programmer, or The Prompt?
An interactive courtroom-style session on AI security, where audiences judge responsibility for five real CVEs in AI tools. It analyzes root causes, evolving vulnerabilities, and the “blame game,” arguing that securing agentic AI requires shared responsibility among developers, providers, and users.
talk.summaryAiDisclaimer
Makan SepehrifarProductboard
talkDetail.whenAndWhere
Thursday, October 8, 12:55-13:35
TBA 8
talks.roomOccupancytalks.noOccupancyInfo
This interactive session transforms AI security education into an engaging courtroom style debate. We'll present five critical vulnerabilities affecting modern AI development tools, and after each case study, the audience becomes the jury—voting on who's responsible: the careless user, the negligent developer, or the inadequate service provider.
We'll dissect real CVEs including GitHub Camo Leak (CVE-2025-59145), Ollama RCE (CVE-2024-37032), AWS MCP Server Shell Injection (CVE-2025-5277), Anthropic MCP Inspector DNS Rebinding (CVE-2025-49596 ) and Microsoft Copilot's zero-click data exfiltration (CVE-2025-32711). Each case reveals technical root causes through collaborative analysis. The List of CVEs will be usually update based on the new related CVEs as well.
The conclusion challenges the "blame game" itself: these vulnerabilities expose fundamental architectural weaknesses in agentic AI systems where traditional security models fail. We will talk about Mental Model Gap and how this phenomenon has been seen before in the technological leaps before. We'll establish that securing AI tools demands a shared responsibility framework developers must code defensively, providers must architect securely, and users must understand AI-specific risks.
We'll dissect real CVEs including GitHub Camo Leak (CVE-2025-59145), Ollama RCE (CVE-2024-37032), AWS MCP Server Shell Injection (CVE-2025-5277), Anthropic MCP Inspector DNS Rebinding (CVE-2025-49596 ) and Microsoft Copilot's zero-click data exfiltration (CVE-2025-32711). Each case reveals technical root causes through collaborative analysis. The List of CVEs will be usually update based on the new related CVEs as well.
The conclusion challenges the "blame game" itself: these vulnerabilities expose fundamental architectural weaknesses in agentic AI systems where traditional security models fail. We will talk about Mental Model Gap and how this phenomenon has been seen before in the technological leaps before. We'll establish that securing AI tools demands a shared responsibility framework developers must code defensively, providers must architect securely, and users must understand AI-specific risks.
Makan Sepehrifar
Makan Sepehrifar is a software architect who is interested in how people think and behave in organizations. He sees software architecture as a craft and believes that every challenge has both a technical side and a social side, and both need attention at the same time. Coming from a strong developer background, he has led many teams across different domains and organizations, always working to connect technology with human needs and team dynamics.
He is passionate about the future of Developer Experience and how Generative AI and Agentic AI are reshaping the way developers work. Makan focuses on how these technologies can improve developer satisfaction and help teams become more adaptive and effective. He is a public speaker on Developer Experience practices and AI in this space, and he has been part of several Generative AI adoption teams in different companies, helping them use AI in a practical and effective way.
He is passionate about the future of Developer Experience and how Generative AI and Agentic AI are reshaping the way developers work. Makan focuses on how these technologies can improve developer satisfaction and help teams become more adaptive and effective. He is a public speaker on Developer Experience practices and AI in this space, and he has been part of several Generative AI adoption teams in different companies, helping them use AI in a practical and effective way.