AI Misidentification of Argentine Activist Leads to Death Threats
Google's AI search system incorrectly identified Argentine activist Flavia Broffoni as the person in a video from Ceuta, Spain, based on facial recognition. The misidentification went viral and triggered threats of violence and death against Broffoni, who was thousands of kilometres away in Argentina at the time.
What happened
A video of an unidentified woman in Ceuta, Spain circulated in WhatsApp groups alongside claims that she was distributing pepper spray to Moroccan migrants. When a user queried Google's AI system with a cropped image of the woman's face and the search term 'activist,' the AI incorrectly identified her as Flavia Broffoni, an Argentine activist and political scientist.
The false identification went viral on social media and across multiple Instagram influencer accounts with hundreds of followers. Broffoni was located in Epuyén, Chubut, Argentina, over 10,000 kilometres away from Ceuta.
Reported impact
Starting on 26 August, Broffoni received an email from an unknown woman in Ceuta warning her that groups were organizing to assault her and advising her not to walk alone. She subsequently received violent messages and death threats on social media.
Comments on the viral posts included calls for 'express justice,' threats to harm her, and incitement to sexual violence. Broffoni stated that comments urged violence against the person she was supposedly identified as, including calls to throw her into the sea.
Index analysis
The AI system's misidentification was facilitated by facial recognition search and a single contextual keyword ('activist') that did not constrain the search geographically or temporally. The system generated a confident identification without apparent safeguards against false matches.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.