OpenAI and Google push new AI models easily tricked into making nudes, researchers say

Source à consulter

Researchers have discovered that the new AI models developed by OpenAI and Google can be easily manipulated to create explicit content. This discovery was announced in a publication on Ars Technica AI.

The researchers identified vulnerabilities in the architectures of the models, which allow attackers to manipulate them to create explicit content. This raises concerns about the security and reliability of these models.

This discovery has significant implications for creators and designers, who must now consider the possibility of manipulation of these models. This could lead to changes in how they design and use these models.

Titre et date issus du flux; image déclarée par la page source, réponse image vérifiée. Synthèse à relire.