World Congress 2025 Aug 20, 2025 Session details

Beyond the Hype: Building Trustworthy and Reliable LLM Applications with Guardrails

Alex Soto

Securing your LLM prompt layer is only half the battle. Deploy robust guardrails to protect RAG pipelines and block prompt injections without crippling system latency.

Pause
Mute Enter Fullscreen
#1 about 2 min

Understanding LLMs as vulnerable software services

Treating LLMs as inference services exposed via REST APIs exposes them to traditional and novel security vulnerabilities.

#2 about 3 min

Categorizing attacks and introducing guardrail defense mechanisms

Deploying input and output guardrails defends against availability, integrity, privacy, and abuse threats while balancing performance costs.

#3 about 3 min

Preventing availability breakdowns through resource limits

Implementing token counting and no-refusal metrics prevents denial of service and detects boundary-testing behavior.

#4 about 6 min

Mitigating integrity violations in user interactions

Applying filters for gibberish, language consistency, malicious URLs, response lengths, and banned topics maintains reliable model interactions.

#5 about 3 min

Protecting systems from prompt injection and sentiment attacks

Guarding against prompt manipulation and negative sentiment inputs prevents unintended model overrides and resource starvation.

#6 about 3 min

Securing privacy with text filtering and anonymization tools

Masking sensitive data and secrets using dedicated anonymizers ensures compliance and safe external data handling.

#7 about 5 min

Blocking abuse through toxicity rules and jailbreak detection

Configuring ban lists and specialized categorization blocks illicit code execution, competitor mentions, toxic content, and do-anything-now exploits.

#8 about 5 min

Implementing Java interceptors for input and output guardrails

Utilizing an interceptor pattern enables applications to programmatically evaluate API payloads against constraint models.

#9 about 4 min

Addressing knowledge base threats in RAG architectures

Securing vector stores against data poisoning ensures accurate retrieval operations and protects underlying operational data structures.

Matching moments

2:47 min

Leveraging older LLMs defensively for vulnerability hunting

Adrian Mouat Adrian Mouat · WWC Europe 2026

2:15 min

Establishing guardrails and infrastructure for AI models

Alexandre Guenoun Alexandre Guenoun +3 · WWC Europe 2026

1:20 min

Utilizing industry threat models for AI security

Balázs Kiss · WWC 2023

3:09 min

Implementing regex, classifiers, and LLM-as-a-judge guardrails

Cansu Kavili Örnek Cansu Kavili Örnek · WWC Europe 2026

4:15 min

Security integration and AI skepticism in developer tooling

Chris Heilmann +2 · LIVE

5:25 min

Addressing core challenges in large language model deployments

Vijay Krishan Gupta +1 · LIVE

Upcoming sessions on this topic

Open session

World Congress 2026 North America

Red Teaming Your LLM App -- A Hands-On Threat Model You Can Reuse

Saloni Garg

Senior ML Engineer at Adobe

Saloni Garg
Open session

World Congress 2026 North America

No Single Model to Rule Them All: Building Resilient AI Agents Across Open & Closed LLMs

Emmanuel Acheampong

Senior Manager Developer Relations at Crusoe AI

Emmanuel Acheampong
Open session

World Congress 2026 North America

SecurePrompt: Building a Pre-Flight Security Layer for Agentic AI

Ravi Sastry Kadali

AI/ML Engineer at General Motors

Ravi Sastry Kadali
Open session

World Congress 2026 North America

The Things Your AI Isn't Telling You

Desmond Lamptey

Lead Software Engineer @ Capital One

Desmond Lamptey
Open session

World Congress 2026 North America

Understanding LLM Architectures: Inside the Design of Modern Models

Jofia Jose Prakash

Enterprise AI Architect at American Chemical Society

Jofia Jose Prakash
Open session

World Congress 2026 North America

Who Tests the AI? Building Trustworthy AI Systems at Enterprise Scale

Him Raj Singh

PayPal, Manager, Software Engineer

Him Raj Singh