The organization has taken down LAION-5B in response to a Stanford report that found links to illegal content. However, concerns remain about the potential misuse of such data, including the creation of explicit deepfakes and revealing sensitive information. In 2022, an artist found her image from her private medical records in the LAION dataset. The Brazilian legislature is considering laws to regulate deepfake creation, and in the US, representative Alexandria Ocasio-Cortez has proposed the DEFIANCE Act, which would allow people to sue if a nonconsensual deepfake in their likeness had been made.
Key takeaways:
- Over 170 images and personal details of children from Brazil have been used without consent to train AI, according to a report from Human Rights Watch.
- The data was included in LAION-5B, a popular dataset for AI startups, and was scraped from content posted as far back as the mid-1990s.
- LAION-5B has been taken down in response to a Stanford report that found links to illegal content, and the organization is working with various groups to remove all known references to illegal content.
- There are concerns that the database could reveal potentially sensitive information, such as locations or medical data, and that children's photos could be stolen and weaponized against them.