YouTube to introduce flagging for AI-mimicked artist voices

Record labels can seek removal of content imitating an artist’s distinctive singing or rapping

YouTube’s latest guidelines empower record companies to seek the removal of songs utilizing AI-generated versions of artists’ voices. The platform is rolling out a tool for music labels to flag content mimicking an artist’s distinctive singing or rapping voice. The surge in generative AI, capable of creating convincing text, images, and voice from human prompts, has given rise to the unintended consequence of counterfeit AI-generated music.

Heart on My Sleeve, a notable instance, showcases AI-generated vocals claiming to be Drake and the Weeknd. Universal Music Group, representing both artists, condemned the song for “infringing content created with generative AI,” leading to its removal from streaming services. Despite this, the song remains accessible to listeners on YouTube.

In a blog post, the Google-owned platform announced its plans to test the new controls with a specific set of labels and distributors before a broader release. This chosen group is actively involved in undisclosed “early AI music experiments,” utilizing generative AI tools for content creation. Additionally, YouTube will enhance its privacy complaint process to enable users to report deepfakes.

The platform stated, “We will enable users to request the removal of AI-generated or other synthetic content that replicates an identifiable individual, encompassing their face or voice.” It clarified that parodic content or deepfakes featuring public figures and well-known personalities may not be taken down.

“In evaluating these requests, we’ll consider various factors,” mentioned YouTube in a blog post by Jennifer Flannery O’Connor and Emily Moxley, VPs of Product Management at the company. Additionally, YouTube will mandate creators to disclose the creation of realistic-looking “manipulated or synthetic” content, which includes AI-generated material. Upon uploading content, creators will have the option to flag synthetic footage. Persistent violations of these guidelines could lead to content removal or suspension of advertising payments, as per YouTube.

YouTube emphasized the significance of its new AI guidelines, particularly in scenarios discussing sensitive subjects like elections, ongoing conflicts, public health crises, or public officials.

When a label is flagged for AI-generated content, it will be incorporated into a video’s description panel. Notably, content addressing sensitive topics will receive a more conspicuous label. YouTube clarified that AI-generated material violating existing content guidelines, such as synthetically created violent videos aiming to elicit shock or disgust, will be entirely removed.

The preceding week, Meta, the parent company overseeing Facebook and Instagram, declared that political advertisers must acknowledge the utilization of AI in their platform ads. Meta specified that advertisers are obligated to reveal instances where AI is employed to depict a genuine individual engaging in actions or utterances they did not genuinely perform or say, encompassing image, video, or audio content.

In the previous month, the UK government issued a cautionary statement, highlighting the potential contribution of deepfakes to the “degradation of the information environment.” It underscored concerns about a possible erosion of public trust in “authentic information, institutions, and civic processes such as elections.”

Recent reports revealed the circulation of fabricated audio clips featuring Sadiq Khan, the Mayor of London, purportedly downplaying the significance of Armistice Day on social media.