Hidden instructions in a court filing
A court in Brazil identified an attempt to mislead AI systems through a prompt injection embedded in a filing submitted by lawyers, and imposed sanctions on those involved. The case drew wide attention on Hacker News.
What is prompt injection?
Prompt injection is a technique in which hidden instructions are placed inside a text so that an AI model processes it differently than intended. Here, the apparent goal was to influence how an AI tool handling the filing would interpret its contents.
Why it matters
As AI tools become more common in judicial systems, instructions hidden inside documents threaten the integrity of court processes. The ruling is an early example showing that such attempts can carry legal consequences.
Limited information
The source does not specify the court's location, the date, or the exact nature of the sanctions; the report was shared at headline level only.
