OOmnistu

Blog / AI Development

AI Development

Prompt injection explained for product teams: what it is and what reduces the damage

Understand how untrusted content can redirect an AI feature and why permission boundaries and independent validation matter.

OOmnistu
Reviewed 2026-11-16 · 3 min read

If your app lets a language model read text from users, websites, files or a database, that text can change what the model does. OWASP lists this as LLM01:2025 Prompt Injection, the first risk in its Top 10 for LLM Applications. We read the project's page on it in full on 15 November 2026. We did not read the references it links to. This article summarises one source and does not claim any defence is complete.

What it is

Prompt injection happens when inputs alter the model's behaviour or output in unintended ways. OWASP says the inputs need not be readable by humans, as long as the model parses them. It can lead the model to break guidelines, produce harmful content, enable unauthorised access or influence critical decisions. The page says retrieval-augmented generation (RAG) and fine-tuning make outputs more relevant but, per research it cites, do not fully remove the problem. It treats jailbreaking as a form of prompt injection where the model is made to disregard its safety rules entirely.

Direct injection comes from the user's own prompt, intentional or accidental. Indirect injection comes from external content the model reads, such as websites or files, again intentional or not.

What can go wrong

OWASP lists: disclosure of sensitive information, leaking system prompts or infrastructure details, biased or wrong output, unauthorised access to functions the model can call, running commands in connected systems, and manipulated decisions. Severity depends on the business context and on how much agency the model has. Multimodal models add risk: instructions can be hidden in an image that accompanies harmless text.

OWASP's mitigations

The page says it is unclear whether foolproof prevention exists, because of how models work, and offers measures that reduce impact:

  1. Constrain model behaviour in the system prompt: role, capabilities, limits, and ignoring attempts to change core instructions.
  2. Define and validate output formats, using ordinary code to check them.
  3. Filter inputs and outputs with rules for sensitive categories, semantic filters and string checks. It mentions evaluating responses with the RAG Triad: context relevance, groundedness and question-answer relevance.
  4. Least privilege. Give the application its own API tokens and handle those functions in code, not through the model.
  5. Human approval for high-risk actions.
  6. Segregate and label external content so untrusted text has less influence.
  7. Adversarial testing, treating the model as an untrusted user to test trust boundaries.

Note the pattern: steps 4 and 5 do not try to make the model immune. They limit what a fooled model can do. See our article on excessive agency for the same idea.

Scenarios from the page

ONE INDIRECT-INJECTION SCENARIO FROM THE OWASP PAGE 1. The user asks "Summarise this page" - an ordinary request 2. The page carries a hidden instruction invisible to the reader, parsed by the model 3. The model follows it adds an image link to its answer, pointing at the attacker's server 4. The image loads and the conversation reaches the attacker with the request ONE INDIRECT-INJECTION SCENARIO FROM THE OWASP PAGE 1. The user asks "Summarise this page" - an ordinary request 2. The page carries a hidden instruction invisible to the reader, parsed by the model 3. The model follows it adds an image link pointing at the attacker's server 4. The image loads and the conversation reaches the attacker with the request
One of the indirect-injection scenarios on the OWASP LLM01 page: a summarised webpage contains hidden instructions that make the model add an image link that leaks the conversation. The mitigations listed above - least privilege, segregating and labelling external content, human approval for high-risk actions - limit what a fooled model can do; the page does not claim any defence is complete, and neither do we.

Questions for a team building an AI feature

Limits

One OWASP page, read in full; its references (papers and blog posts) were not read. We make no claim about any model, vendor or filter and give no statistics.

By Omnistu.

Sources

Previous
Treat AI model output as untrusted input: improper output handling explained