Glossary: AI Security and Guardrails
Improper Output Handling
Also written: LLM10
An OWASP risk in which model output is shown or passed on without safe handling, for example a reply's image link leaking data. Sanitize, parameterize and validate.
Glossary: AI Security and Guardrails
Also written: LLM10
An OWASP risk in which model output is shown or passed on without safe handling, for example a reply's image link leaking data. Sanitize, parameterize and validate.