Autonomous Software and Agentic Security

An AI Security Engineers meetup on what autonomous agents do to an attack surface. The first talk, from Gerald Crescione, who runs Snyk's AI Security Engineers community across more than thirty chapters, argues that software has become autonomous while security teams have not, and looks at where traditional AppSec models stop working against probabilistic systems. The second goes concrete: how a prompt injection hidden in an MCP tool description or a GitHub issue can hijack an agent, and why a container was never a security boundary — through to cluster control. Hosted by WeScale at 34 boulevard Haussmann. Free, in French, open registration. Why go: this is the specific attack surface most teams shipping agents have not yet threat-modelled, treated by practitioners rather than vendors. Best for security engineers and anyone putting agents near real systems.