← All articles

// ai security · 7 min

Securing AI Chatbots: A Practical LLM Security Checklist

Every company is shipping an AI assistant. Very few have tested it like an attacker would. LLM applications introduce failure modes that traditional pentests never covered — and the OWASP Top 10 for LLM Applications now formalises them.

The checklist

1. Prompt injection resistance

Can user input — or content the model reads from documents, emails, or web pages — override your system instructions? Indirect injection through retrieved content is the attack most teams miss.

2. Sensitive data exposure

Does the model have access to more data than the current user should see? Test cross-tenant and cross-role leakage explicitly, with adversarial phrasing.

3. Excessive agency

If the model can call tools — send emails, query databases, execute actions — what is the blast radius when it is manipulated? Every tool call needs authorisation checks independent of the model's judgement.

4. Output handling

Model output rendered into HTML, SQL, or shell commands is untrusted input. Treat it with the same sanitisation discipline as user input, because an attacker controls it indirectly.

5. Guardrail bypass testing

Content filters and system prompts are speed bumps, not walls. Red-team them with role-play framing, encoding tricks, and multi-turn setups before someone on the internet does.

Before launch

Run a dedicated AI/LLM security assessment covering the full OWASP LLM Top 10 against your actual application — not the base model. The base model's safety is the vendor's job; your integration is yours.