Can YouTube control third-party AI training from user videos?

Question

Accepted Answer

YouTube's Control Over Third-Party AI Training from User Videos is Limited

YouTube's current data-sharing policy indicates that while creators can opt in to share their videos for training AI models, the platform has limited control over how third-party companies utilize this content. This situation has raised concerns for public figures, including Bollywood stars, regarding the potential misuse of their likeness and intellectual property in AI-generated content. The legal challenges brought forth highlight the complexities of personality rights in the digital age and the difficulties in preventing the proliferation of infringing material.
YouTube's Policy on Creator Opt-In for AI Training

YouTube's policy allows creators to voluntarily share their uploaded videos for the purpose of training other AI platforms. This opt-in mechanism means that the decision to contribute content rests with the individual creator. However, the platform explicitly states that it "can’t control what a third-party company does" with the shared videos once they are made available for training. This statement underscores a significant gap in YouTube's oversight of how this user-consented data is subsequently used by external AI developers.
Legal Challenges Regarding AI Training Data and Personality Rights

High-profile individuals, such as Bollywood actors Abhishek Bachchan and Aishwarya Rai Bachchan, have initiated legal proceedings to protect their intellectual property and personality rights from AI-driven infringements. Their lawsuits target not only the unauthorized creation and distribution of AI-manipulated content but also seek to compel Google to implement safeguards against such videos being used to train other AI models. This legal action reflects a broader trend of celebrities asserting their rights in courts as India lacks explicit legal protections for personality rights akin to those in some US states.
The Bachchans' Lawsuit and Court Orders

The Bachchans' legal challenge includes claims against Google and others for damages amounting to US$450,000 (RM1.9 million) and requests for a permanent injunction against the exploitation of their image and voice. In early September, a judge ordered the takedown of 518 specific website links and posts identified by the actors, acknowledging the financial harm and damage to their dignity and goodwill. Despite these orders, investigations have revealed the continued presence of AI-manipulated clips and depictions of actors on YouTube, similar to those cited in the legal filings.
Concerns Over Proliferation of Misleading Content

A core argument presented in the Bachchans' filings is that the use of biased or infringing content to train AI models risks multiplying instances of misinformation. Once uploaded to YouTube and viewed by the public, such content can then be incorporated into AI models, leading to its further and potentially wider dissemination. This concern is amplified by the discovery of channels dedicated to "AI-generated Bollywood love stories," which utilize AI tools to create content that, while presented as entertainment, may still rely on the likenesses and personas of real individuals without explicit consent for such applications.

Reading history

Hide →

This history is cleared when the page is closed.

Can YouTube control third-party AI training from user videos?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends