Refine speaker talk details in online event markdown

Updated speaker talk titles and synopses for clarity and formatting consistency.

Author: MatOwasp

events/2026-02-13-online.md

@@ -15,10 +15,10 @@ Starting at 4 PM CET
 
 ### <img src="https://github.com/OWASP/www-chapter-italy/blob/master/assets/images/speakers/simone-curzi.jpeg?raw=true" alt="Marco Morana" width="34" height="34" style="border-radius:50%; vertical-align:middle; margin-right:8px;" /> Simone Curzi - Microsoft 
 
-**Talk:** *Am I Secure Enough? Rethinking Security as an Investment with Quantitative Risk Analysis*
+**Talk:** 
+*Am I Secure Enough? Rethinking Security as an Investment with Quantitative Risk Analysis*
 
-**Synopsis**  
-Security is often treated as a non-negotiable cost: we must be secure, no matter what. In the real world, resources are finite and security decisions have a cost, an operational impact, and a measurable effect on risk.
+**Synopsis**: Security is often treated as a non-negotiable cost: we must be secure, no matter what. In the real world, resources are finite and security decisions have a cost, an operational impact, and a measurable effect on risk.
 
 This talk challenges the idea that one should not ask “Am I secure enough?” or “Which mitigations should I focus on?”. On the contrary, these are the most important questions security professionals should be able to answer.
 
@@ -32,10 +32,10 @@ The result is not just better prioritization, but a shared language that allows
 
 ### <img src="https://github.com/OWASP/www-chapter-italy/blob/master/assets/images/trainers/Krish.jpg?raw=true" alt="Krishnendu Dasgupta" width="34" height="34" style="border-radius:50%; vertical-align:middle; margin-right:8px;" /> Krishnendu Dasgupta - AXONVERTEX AI  
 
-**Talk (draft):** *Your Agents Have Power — Do They Have Guardrails? Engineering Agent Security, Privacy, and Policy in Decentralized Local Compute Setups*
+**Talk:** 
+*Your Agents Have Power — Do They Have Guardrails? Engineering Agent Security, Privacy, and Policy in Decentralized Local Compute Setups*
 
-**Synopsis** 
-As AI agents gain autonomy to plan, call tools, access data, and execute actions, they introduce new security and privacy risks—especially in decentralized, locally hosted environments. This session shows how to engineer enforceable guardrails for agentic systems using open-source components and local models, without relying on centralized cloud controls. Attendees will learn how to define risk boundaries, threat-model agent workflows, apply least-privilege tool access, enforce policy-as-code, add privacy safeguards, enable audit-ready observability, and continuously test agent behaviors. The talk concludes with a preview of swarm agents in trustless compute, highlighting what’s next for secure, decentralized agent architectures.
+**Synopsis**: As AI agents gain autonomy to plan, call tools, access data, and execute actions, they introduce new security and privacy risks—especially in decentralized, locally hosted environments. This session shows how to engineer enforceable guardrails for agentic systems using open-source components and local models, without relying on centralized cloud controls. Attendees will learn how to define risk boundaries, threat-model agent workflows, apply least-privilege tool access, enforce policy-as-code, add privacy safeguards, enable audit-ready observability, and continuously test agent behaviors. The talk concludes with a preview of swarm agents in trustless compute, highlighting what’s next for secure, decentralized agent architectures.
 
 
 Back to the [OWASP-Italy Chapter](https://owasp.org/www-chapter-italy)