Brazilian Court Sanctions Lawyers for Prompt Injection in Court Filing

A Brazilian labor court censured and fined two lawyers after finding they intentionally embedded a hidden prompt-injection command within a court filing. The case highlights growing concerns over the security and integrity of artificial intelligence systems used within legal proceedings.

• Two law firms were fined 10% of the claim's value, totaling approximately $16,000.
• The petition contained hidden text in white font on a white background, containing a prompt-injection command.
• The tampering was detected by Galileu, a generative AI tool utilized nationally by Brazil's National Justice Council.
• The court ruled the action violated procedural good-faith obligations and undermined the dignity of justice.

Discovery of the Hidden Prompt Injection

The manipulation was uncovered while a judge was reviewing an employment petition through Galileu, a generative AI system deployed across Brazilian courts. The system automatically flagged an anomaly and blocked the processing of suspicious content after identifying text formatted in white font against a white background.

Once the font color was adjusted, the concealed instruction became fully visible to court officials. The prompt directed any AI processing it to handle the petition superficially and avoid challenging the presented documents, aiming to generate a biased judgment to the detriment of the opposing party.

Legal Basis and Judicial Sanctions

The court based its sanctions on the Brazilian Civil Procedure Code, citing articles that mandate procedural good-faith, truthfulness, and respect for judicial institutions. The code strictly prohibits parties from obstructing justice or introducing unauthorized alterations into legal proceedings.

Because the original employment claim was valued at 842,000 Brazilian reals, the resulting 10-percent fine amounted to about $16,000. The judge emphasized that such deceptive tactics severely threaten the integrity of judicial processes.

Implications for AI Governance in Legal Systems

As courts worldwide increasingly adopt artificial intelligence for document management and case summaries, this ruling sets a crucial legal precedent. It confirms that the standard of truthful filing applies equally to machine-readable digital content.

Furthermore, the automated detection by the Galileu system underscores the necessity of robust enterprise AI governance, emphasizing that modern legal and corporate tools must continuously monitor for adversarial prompt injection attempts.