The platform is rolling out a new feature that permits them to cut down on the quantity of AI-made videos appearing in their feeds. This move comes as the service revealed that it now features more than 1 billion AI-generated clips.
Trial and Implementation
The update is being tested over the next few weeks before a global release. This shift aligns with a surge in AI content, spurred by new AI video creators.
AI Content on Other Platforms
It has been noted that approximately 10% of the fastest-growing content creators on another major platform primarily display clips produced artificially. A large amount of this content is considered mass-produced AI material, a term for poor-quality, factory-style content that is typically nonsensical or surreal.
“We know from our community that many people appreciate content produced via AI tools, ranging from computer-generated images to informative clips, and we want to give people the power to increase or decrease of that, based on their own preferences,” said a platform representative.
Scale of AI Content
The platform announced that there are currently 1.3bn videos on its service identified as AI-generated. In total, more than 100 m media items are uploaded to the app each day, making clips made by AI a limited portion of its total content library.
How to Adjust Settings
Members will be in a position to reduce or increase the amount of AI content they watch by using the topic control feature in the platform and choosing AI-made media. Further subjects that currently be controlled under the setting include politics, style, cosmetics, and movement.
Content Labeling and Policies
Its rules oblige creators to identify “realistic” AI-generated content, and it prohibits dangerous deepfakes of public figures or crisis events. Any realistic artificial content that is without a tag can be deleted under its community policies.
Additionally, the platform will henceforth apply an artificially created watermark to content created with its own AI tools, or detected by an collaborative effort for identifying content made by AI. This measure is designed to help prevent users from attempting to evade the labeling process.
AI Literacy Programs
The company is also introducing a two million dollar educational fund for experts and organizations to create informative material about using AI responsibly. This program will benefit non-profit groups specializing in training coding and technology.
Content Safety Approach
The platform has justified its content review process, following discussion over intentions to make redundant a large number of content moderators in the United Kingdom. It is planning to reduce 439 jobs in its trust and safety division in London, with worker representatives and digital protection specialists voicing alarm that staff are being displaced by machine-led review.
A senior executive clarified that people-led review would remain essential, but AI can protect employees by eliminating the most harmful content prior to it has to be seen by staff. TikTok stated a notable drop in disturbing and violent content seen by human moderators over the past year due to the use of its machine learning processes.
It is vital to balance humans and technology to ensure user safety,” the executive stated. “We’ll have staff as part of that process, but we also think it’s extremely critical to limit contact with unsafe media as rapidly, which is the area in which a lot of the machine support is making a difference process content quickly.”
