Skip to main content
NOPE
Medium reported severity Regulatory Action

Pará labour court fines two lawyers about R$84,000 over hidden instruction aimed at the court's Galileu AI; the lawyers deny manipulating the decision

In May 2026 a judge of the 3rd Labour Court of Parauapebas, in the Brazilian state of Pará, fined two lawyers about R$84,000 after text in white font on a white background was found in a document they filed for a worker in a labour claim. According to the sentence as reported by G1, the hidden text was detected while the judgment was being drafted with Galileu, a generative AI tool the labour courts use to help judges and staff draft sentences, and its evident intention was to make the AI produce a superficial defence or a compromised draft judgment. The judge called the technique prompt injection. The lawyers said they disagreed with the decision and that there was never any command to manipulate the judge's decision, describing the text as an attempt to protect their client from the AI; one of them said she did not work on labour cases. In October 2026 the regional labour court announced that a Connecticut Superior Court judge had cited the decision when sanctioning a litigant for misuse of AI.

AI system
Galileu
Occurred
May 2026
Reported
15 October 2026
Event location
Brazil
What the AI did
Relation unknown
Reported harm
Legal HarmFinancial LossProfessional Harm
Whose AI use
An institution’s AI use
Setting
Justice · Work
Evidence
AI involvement supported · Causal attribution disputed · 3 sources, 2 underlying accounts
5 claims: 5 reported. 4 open questions
People reported harmed
2 people

AI system as recorded: Galileu, a generative AI tool developed by the 4th Region Labour Court with the Supreme Federal Court to help judges and staff draft sentences, used by the 8th Region Labour Court; the hidden text was aimed at it and was detected while the judgment was drafted with it

What Happened

The hidden text. In a labour claim brought by a worker before the 3rd Labour Court of Parauapebas, which began in July 2025, the worker's lawyers filed a document containing text in white font on a white background, invisible to human readers. G1 reports that the text was detected while the judgment was being drafted with Galileu, the generative AI tool the labour courts use to help draft sentences. According to the 8th Region Labour Court, the aim was to force a superficial reading of the petition without challenges to the documents.

The sentence. In May 2026 the judge classified the technique as prompt injection. G1 quotes the sentence as saying the intention was evident: to make the AI system generate a superficial defence or a compromised draft judgment. The judge fined the two lawyers about R$84,000.

The lawyers' response. In notes sent in May, the lawyers said they disagreed with the decision, that there was never any command to manipulate the judge's or any official's decision, that the text concerned the defence and was meant to protect their client from the AI itself, and that they expected the decision to be reversed. One of them said she worked only in civil and social-security law, took no part in labour cases and had not seen the case file before the sentence.

International citation. On 14 October 2026 the 8th Region Labour Court announced that a Connecticut Superior Court judge had cited the decision, as an example of another court treating materially identical conduct as an offence against the integrity of proceedings, when he sanctioned a litigant for misuse of AI.

Reported harm

According to reports of the sentence, two lawyers were fined about R$84,000 by a labour court that found they had hidden an instruction aimed at the court's generative AI drafting tool in a filed document; the lawyers dispute the finding.

Outcome

Unknown

The 3rd Labour Court of Parauapebas (8th Region Labour Court) fined the two lawyers about R$84,000 (R$84,200 according to Metrópoles) in May 2026. The lawyers said they disagreed and expected the decision to be reversed; whether they appealed and with what result is not reported. On 14 October 2026 the 8th Region Labour Court announced that Judge Walter M. Spader Jr. of the Connecticut Superior Court had cited the decision when sanctioning a litigant.

What remains unknown

  • The sentence's exact date and case number; the reports give only May 2026.
  • Whether Galileu itself flagged the hidden text, as G1 states, or a person noticed it during AI-assisted drafting, as the sentence's wording allows.
  • Whether the lawyers appealed and whether the fine stands.
  • Any consequence for the worker's claim.

What the evidence supports

AI involvement: supported. G1 reports that the hidden white-on-white text was detected while the judgment was being drafted with Galileu, a generative AI tool the labour courts use to help draft sentences, and quotes the sentence as finding that the text was meant to make the AI system produce a superficial defence or a compromised draft judgment. The harm is the court's response to the hidden instruction aimed at generative AI (the fine on the two lawyers), so no relation describing an AI system's own action toward the lawyers is established (unknown; consequential response). G1 states that the hidden text was identified by Galileu itself, while the sentence as quoted says only that it was detected during drafting via Galileu.

5 claims: 5 reported. What the statuses mean

Reported In May 2026 a judge of the 3rd Labour Court of Parauapebas fined two lawyers about R$84,000 after a hidden instruction in white font was found in a document filed in a labour case, aimed at the court's Galileu AI.

Causal attribution. Reports of the sentence and of the court's announcement; the sentence itself was not read.

  • g1.globo.com(opens in new tab) supports · Portuguese
    'A punição ocorreu após a identificação de um comando oculto com fonte branca em um documento trabalhista. O objetivo era enganar a IA Galileu, usada pelo tribunal.'; 'A decisão no Pará foi tomada em maio de 2026'
  • metropoles.com(opens in new tab) supports · Portuguese
    'Em maio, o juiz brasileiro multou em R$ 84,2 mil duas advogadas do Pará após elas tentarem enganar a inteligência artificial Galileu'
Reported According to the sentence as reported by G1, the text was detected while the judgment was being drafted with Galileu, and its evident intention was to make the AI system produce a superficial defence or a compromised draft judgment.

Causal attribution. The sentence's finding as quoted by G1; the lawyers dispute it (c4).

  • g1.globo.com(opens in new tab) supports · Portuguese
    'foi detectado um texto em fonte branca sobre fundo branco'; 'A intenção era evidente: fazer com que o sistema de IA utilizado gerasse uma contestação superficial ou uma minuta de sentença comprometida'
Reported Galileu is a generative AI tool developed by the 4th Region Labour Court with the Supreme Federal Court to help judges and staff draft sentences.

Causal attribution. Description from the court announcement as relayed by DOL.

  • dol.com.br(opens in new tab) supports · Portuguese
    'Desenvolvido pelo TRT-4 em parceria com o Supremo Tribunal Federal (STF), o Galileu é uma ferramenta de inteligência artificial generativa que auxilia magistrados e servidores na elaboração de minutas de sentenças'
Reported The lawyers said they disagreed with the decision and that there was never any command to manipulate the judge's decision, describing the text as protection for their client against the AI; one said she took no part in labour cases.

Causal attribution. The lawyers' own statements to G1 in May 2026.

  • g1.globo.com(opens in new tab) supports · Portuguese
    'jamais existiu qualquer comando para manipular a decisão do Magistrado ou de qualquer outro servidor'; 'uma tentativa de proteger o nosso cliente da própria IA e nada mais que isso'; 'não detinha qualquer participação em processos da seara trabalhista'
Reported A Connecticut Superior Court judge cited the Pará decision when sanctioning a litigant for misuse of AI, as the 8th Region Labour Court announced in October 2026.

Causal attribution. The court's announcement as relayed; the Connecticut decision was not read.

  • metropoles.com(opens in new tab) supports · Portuguese
    'foi usada como referência pelo juiz Walter M. Spader Jr., da Corte Superior de Connecticut, nos Estados Unidos'
  • g1.globo.com(opens in new tab) supports · Portuguese
    'A informação foi divulgada pelo Tribunal Regional do Trabalho da 8ª Região (TRT-8) nesta quarta-feira (14).'

Sources

3 sources inspected, from 2 underlying accounts. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read in Portuguese by the research agent (an AI) on 2026-10-16 (HTTP 200); passages translated by the research agent; no human translation. Applies to s1, s3.

Read in Portuguese by the research agent (an AI) on 2026-10-16 (HTTP 200); passages translated by the research agent; no human translation. Metrópoles relays the 8th Region Labour Court's announcement and quotes the Connecticut decision. Applies to s2.

Event countries: Brazil. Affected-person countries: Brazil. Court countries: Brazil.

The document was filed in, and the fine imposed by, the 3rd Labour Court of Parauapebas, Pará; the lawyers practise in Parauapebas according to G1.

Reviewed for publication 2026-10-16: Published as a consequential-response case: reports of the sentence state that a labour court fined two lawyers for hidden text aimed at the court's generative AI drafting tool, which the judge found was meant to make the AI produce a compromised draft. G1, Metrópoles and DOL were read; the sentence itself was not.

People described

Two lawyers in Parauapebas, Pará, fined in a labour case; they dispute the decision. They are private people and are not named here.

People reported harmed in this case

2 people

0 AI participants · 2 other people harmed

Two people: the two lawyers fined in the sentence. The worker they represented is not counted because no consequence for him is reported.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Tags

#brazil#para#labour-court#prompt-injection#legal-profession#sanction#court-ai-tool

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Pará labour court fines two lawyers about R$84,000 over hidden instruction aimed at the court's Galileu AI; the lawyers deny manipulating the decision. AI incidents. https://nope.net/incidents/2026-para-labour-court-fines-two-lawyers-over-hidden-instruction-aimed-at-court-galileu-ai

BibTeX

@misc{2026_para_labour_court_fines_two_lawyers_over_hidden_instruction_aimed_at_court_galileu_ai,
  title = {Pará labour court fines two lawyers about R\$84,000 over hidden instruction aimed at the court's Galileu AI; the lawyers deny manipulating the decision},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2026-para-labour-court-fines-two-lawyers-over-hidden-instruction-aimed-at-court-galileu-ai}
}

Related cases

Low Unidentified AI drafting tool

Paraíba, Brazil: judge fines a lawyer R$32,800 over hidden AI commands in a petition, G1 reports; state bar later suspends two lawyers in such cases

A judge of the 5th Mixed Court of Sousa, Paraíba, found hidden instructions aimed at artificial intelligence tools spread over seven pages of a lawyer's motion for clarification in a public-competition case in June 2026, including an order to ignore impartiality and a note calling it a test of whether the judge used only AI to decide. The judge fined the lawyer two penalties of R$16,400 (R$32,800 in total) for bad-faith litigation and contempt of the court's dignity, and referred the case to the Paraíba bar (OAB-PB) and the state prosecutors, G1 reports. The defence said the commands were not inserted deliberately and blamed an error in exporting a file prepared with AI support. An appeal judge refused to suspend the fines on 30 June, the trial judge kept them on reconsideration, and in July, G1 reports, the OAB-PB preventively suspended two lawyers after opening procedures into two hidden-command cases, one of them the Sousa case.

Medium Unidentified AI tool

Ninth Circuit fines and suspends two immigration lawyers over briefs with AI-hallucinated citations and denials of AI use; immigration board follows

On 3 June 2026 the US Court of Appeals for the Ninth Circuit fined immigration attorneys Mike Singh Sethi and William Rounds $2,500 each and suspended both from practice before the court for six months. Briefs filed in an asylum petition for review cited two cases that do not exist and attributed quotations to opinions that do not contain them. In a motion to correct, the attorneys called the errors typographical, and at oral argument Rounds twice said AI was not used. In their response to the court's order to show cause, they conceded that the errors were probably the product of unauthorised AI use by the firm's unlicensed brief writers and a failure to check citations. The court said the suspension was owed to their repeated failure of candour. The Board of Immigration Appeals then suspended Sethi from practice before the Board, the immigration courts and the Department of Homeland Security for six months, effective 13 June 2026, and designated that decision as a precedent on 1 October 2026.

Medium Unidentified AI research tool

Alabama Supreme Court dismisses two siblings' family-trust appeal as a sanction after their lawyer admits using AI that produced fake citations

On 24 April 2026 the Supreme Court of Alabama dismissed an appeal by two siblings in a family dispute over living trusts because their lawyer, W. Perry Hall, had filed briefs in which most of the legal authorities were misquoted or did not exist. The court said they appeared to be AI hallucinations, and at a show-cause hearing in February 2026 Hall admitted that many authorities did not exist and that he had improperly used AI to obtain them. The court ordered Hall to pay the opposing party's fees and costs of $17,200 and double appeal costs, barred him from filing in the court without a co-signing attorney and referred him to the Alabama State Bar. It said the dismissal was imposed as a sanction because of the particularly egregious facts. One justice, concurring in part and dissenting in part, objected to dismissing the appeal as a sanction for the lawyer's conduct but would have affirmed the judgment against the clients on the merits.

Medium vLex, Paxton AI and NotebookLM (reported)

New York federal court enters default judgment against a defendant and orders its lawyer to pay $38,877 after false citations it found AI produced

In a federal trademark suit in the U.S. District Court for the Southern District of New York, the court on 5 February 2026 struck a company defendant's motion papers and entered default judgment against it as a sanction on its lawyer, after his filings repeatedly cited nonexistent cases and misattributed quotations. The lawyer said he had found and cite-checked cases with AI tools (vLex and Paxton AI) and used NotebookLM to review a later submission; the court rejected his explanations and found he acted in bad faith. On 9 October 2026 the court described the citations as produced by generative AI and ordered the lawyer to pay $38,877.59 in fees and costs to the firm of a co-defendant's former lawyer, who had reported the false citations to the court.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents