São Paulo, June 10, 2024 (Lusa) — Photographs of Brazilian children and adolescents are being used to create powerful artificial intelligence (AI) tools without the knowledge or consent of those targeted, denounced today, in a statement, Human Rights Watch (HRW).
The organization that works in defense of human rights stated that the photos are taken from the world wide web and inserted into a large set of data that companies use to train their AI tools.
“Children and teenagers should not have to live in fear that their photos could be stolen and used against them. The Government should urgently adopt policies to protect children’s data from AI-driven misuse,” said Hye Jung Han, child rights and technology researcher at HRW.
HRW reported finding that LAION-5B, a dataset used to train popular AI tools and built by scraping much of the Internet, contains links to identifiable photos of Brazilian children.
The names of some children are listed in their captions or in the URL where the image is stored. In many cases, their identities are easily traceable, including information about when and where the child was at the time the photo was taken.
“HRW found 170 photos of children from at least 10 states: Alagoas, Bahia, Ceará, Mato Grosso do Sul, Minas Gerais, Paraná, Rio de Janeiro, Rio Grande do Sul, Santa Catarina and São Paulo This is likely to be a significant underestimation of the total amount of children’s personal data in LAION-5B, as it analyzed less than 0.0001% of the 5.85 billion images. and captions contained in the dataset”, he highlighted.
The investigation found that many of these photos were originally viewed by a few people and appear to have had some measure of privacy previously. It doesn’t seem possible to find them through an ‘online’ search. Therefore, when data is collected and fed into AI systems, these children face more threats to their privacy due to flaws in the technology.
“AI models, including those trained on LAION-5B, are notorious for making private information public; they can reproduce identical copies of the material they were trained on, including medical records and photos of real people. The barriers protection measures established by some companies to prevent the public disclosure of confidential data have been repeatedly breached,” HRW pointed out.
“Bad actors use LAION-trained AI tools to generate explicit images of children from harmless photos, as well as explicit images of child survivors whose sexual abuse images were [guardadas] for LAION-5B”, he added.
In response to HRW, LAION, the German non-profit organization that manages LAION-5B, confirmed that the dataset contained the personal photos of the children found and committed to removing them.
The German organization disputed that AI models trained on LAION-5B could reproduce personal data verbatim. LAION also said that children and their parents are responsible for removing their personal photos from the Internet, arguing that it is the most effective protection against misuse.
HRW highlighted that Brazilian legislators need to propose a ban on the non-consensual use of AI to generate sexually explicit images of people, including children.
“These efforts are urgent and important, but they address just one symptom of a deeper problem: the fact that children’s personal data remains largely unprotected against misuse. As written , Brazil’s General Personal Data Protection Law does not offer sufficient protection for children,” said HRW.
The organization argued that the Brazilian Government needs to strengthen the data protection law, adopting additional and comprehensive safeguards for the privacy of children’s data and recalled that, in April, the National Rights Council of Children and Adolescents (CONANDA) of Brazil published a resolution asking the Council and the Ministry of Human Rights and Citizenship to develop a national policy to protect the rights of children and adolescents in the digital environment within 90 days.
“The new policy should prohibit the use of children’s personal data for AI systems considering the privacy risks involved and the potential for new forms of misuse as the technology evolves,” HRW stressed.
“Generative AI is still a nascent technology, and the associated harms that children are already facing are not inevitable. Protecting children’s data privacy now will help shape developing this technology into one that promotes, rather than violates, children’s rights,” concluded HRW’s Hye Jung Han.
