How it works
Website chatbot security basics
Prompt injection, data leakage, abuse and rate limits — the security questions to ask before putting an AI chatbot on a public website.
In short
The main risks of a public chatbot are prompt injection, leaking information you did not mean to publish, and abuse of the message allowance.
Keep the knowledge base to public content, apply rate limits, and never give the bot access to systems it does not need.
Key facts
Key facts
- Only index content you are happy for anyone to read.
- Rate limits protect against abuse and runaway usage.
- A bot with no system access cannot be tricked into performing actions.
- Prompt injection attempts should be treated as untrusted input, not instructions.
Do not index what you would not publish
The simplest security rule for a website chatbot is that its knowledge base and your public site should contain the same information. Internal pricing sheets, staff notes and draft pages do not belong in it.
Limit what it can do
A chatbot that only reads content and captures enquiries has a very small blast radius. Every integration you add — order lookups, account changes — expands it, so add them deliberately and with verification.
Expect people to test it
- · Someone will ask it to ignore its instructions.
- · Someone will paste a long block of text pretending to be a system message.
- · Someone will spam it to burn your allowance.
- · Grounding, refusal instructions and rate limits handle all three.
The product behind this guide
A chatbot trained on your own website, live today
Defaltam builds AI chatbots that read your public pages and answer with your real prices, services and policies. Support is €79/month, Sales is €149/month, both start with a 5-day free trial, and installation is one script tag.
FAQ
Questions people ask about this
Is it safe to put an AI chatbot on a public website?
Yes, provided its knowledge base contains only content you would publish anyway, it has rate limits, and it has no access to systems beyond answering and capturing enquiries.
What is prompt injection?
A visitor writing text designed to override the bot's instructions. A well-configured bot treats visitor messages as questions, not as instructions about how to behave.
Can someone extract my private data?
Only what you indexed. Keep internal documents out of the knowledge base and there is nothing private to extract.
How is abuse handled?
Message and usage limits per session, which also protect you from runaway costs.
Related
Keep reading.
How AI chatbots actually work →
From the visitor's question to the answer on screen: retrieval, context, the language model and why grounding in your content matters.
How to stop a chatbot from making things up →
Hallucinated prices and invented policies destroy trust. The four controls that keep a website chatbot honest, and how to test for them.
Running a multilingual AI chatbot →
How a modern chatbot answers in the visitor's language from a single-language website, where it goes wrong, and what to translate anyway.
Website chatbots and GDPR: what you need in place →
What data a website chatbot collects, what your privacy policy must say, and how to handle transcripts and consent under GDPR.
Designing a chatbot handoff to a human →
When a bot should stop answering, what it should collect before passing over, and how to make the handoff feel like service rather than failure.
Why most website chatbots fail →
Chatbots fail for four predictable reasons: no clear job, thin content, no handoff and nobody reading the logs. Each has a straightforward fix.
AI chatbot plans and pricing →
Support €79/mo, Sales €149/mo, 5-day free trial.
How to add an AI chatbot to your website →
The ten-minute install guide, platform by platform.