started · updated
OpenAI alerts Brazilian authorities to ChatGPT threats against President Lula
OpenAI has alerted the Brazilian Federal Police (PF) and the Superior Electoral Court (TSE) after identifying conversations on ChatGPT that indicated potential assassination plots against President Luiz Inácio Lula da Silva.
At least two users have been identified. In one prominent case, a user discussed observing the president at a campaign event from a distance of 400 meters and inquired about the legal consequences of killing him. This user also reportedly discussed using a precision rifle, specifically an AR-15 with a scope, shared Nazi symbols, and mentioned escape routes and the use of binoculars.
Other identified messages included references to explosives, potential attacks on medical centers, and the possibility of committing a massacre if the user had sufficient financial resources. OpenAI stated it has a policy of notifying authorities when it identifies dialogues that indicate an “imminent and concrete risk” of physical harm to others. The company has terminated the account of at least one involved user.
The Federal Police have forwarded the documentation to the Ministry of Justice to determine whether to open formal investigations into threats and injury. As of the latest reports, no formal indictments have been issued.
Entities
Andrei Rodrigues · Federal Police · Luiz Inácio Lula da Silva · OpenAI · Polícia Federal · Superior Electoral Court · Tribunal Superior Eleitoral
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 9 SOURCES] OpenAI identified at least two users expressing intentions to attack or threaten President Luiz Inácio Lula da Silva via ChatGPT. painelpolitico.com · www.diariodocentrodomundo.com.br · www.gp1.com.br · jornal24horas.com.br · atarde.com.br · +4 more
- [● 4 SOURCES] The Director-General of the Federal Police forwarded the documentation to the Minister of Justice for evaluation. painelpolitico.com · www.diariodocentrodomundo.com.br · www.gp1.com.br · www.osul.com.br
- [● 2 SOURCES] OpenAI terminated the account of one of the users involved in the threatening conversations. www.gp1.com.br · www.em.com.br
- [○ 1 SOURCE] As of the reports, no formal investigation or indictment has been officially opened against the users. painelpolitico.com
- [● 2 SOURCES] The identified conversations included references to explosives and potential attacks on medical centers. www.gp1.com.br · atarde.com.br
- [● 9 SOURCES] OpenAI notified the Superior Electoral Court (TSE) regarding the threats in accordance with its electoral compliance plan. painelpolitico.com · www.diariodocentrodomundo.com.br · www.gp1.com.br · jornal24horas.com.br · atarde.com.br · +4 more
- [● 8 SOURCES] One user discussed using a precision rifle, specifically an AR-15 with a scope, to target the president. painelpolitico.com · www.gp1.com.br · jornal24horas.com.br · atarde.com.br · www.em.com.br · +3 more
- [● 6 SOURCES] One user shared Nazi symbols and discussed escape routes during the interactions. painelpolitico.com · www.gp1.com.br · jornal24horas.com.br · atarde.com.br · www.em.com.br · +1 more