AI voice cloning used in lawyer impersonation fraud scheme in São Paulo
A criminal organization in São Paulo used AI voice cloning to impersonate lawyers and fraudulently convince victims to transfer funds between October 2025 and April 2026, moving approximately R$ 10 million. Police arrested eight suspects in May 2026 after identifying at least 12 victims.
What happened
A criminal organization used AI technology to clone the voices of real lawyers and fraudulently impersonate them via a virtual phone system. The group contacted victims who had ongoing lawsuits and convinced them to transfer funds by falsely promising release of legal funds, using forged judicial decisions and the cloned voices of actual lawyers whose data was obtained online.
Between October 2025 and April 2026, the gang moved approximately R$ 10 million across São Paulo state and other Brazilian states. On 19 May 2026, a joint task force of São Paulo Civil Police and OAB-SP executed 26 search warrants and arrested at least eight suspects in the capital, though at least ten people were involved overall.
Reported impact
At least 12 victims were identified in cities including São José do Rio Preto, Novo Horizonte, Paulínia and Bauru, with one victim losing R$ 35,000. Police indicated the actual number of victims and total financial loss may be substantially higher as some cases were not formally reported.
Index analysis
The scheme targeted people with active lawsuits by using publicly available online data about court cases to identify both potential victims and the lawyers handling their cases. The organization operated from homes in the capital, suburbs and coastal areas of São Paulo, with members described as lower-middle class youth working entirely online.
Governance controls
The investigation was coordinated by the Rio Preto Regional Police Station and conducted jointly with OAB-SP as a formal operation called 'SP Advocacia Mais Segura'. The lawyers whose voices were cloned were considered victims of the fraud for having their identities and credentials misused.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.