The platform is rolling out a updated setting that allows them to reduce the amount of artificial intelligence-produced videos appearing in their streams. This move comes as the service announced that it currently hosts more than 1 billion AI videos.
Testing and Rollout
The new functionality is undergoing trials over the coming weeks before a full-scale release. This development comes amid a increase in AI content, spurred by emerging video-generation tools.
The Broader AI Trend
Recent reports that nearly 10% of the fastest-growing video streams on a popular video site only show content made by AI. A significant portion of this output is considered “AI slop”, a label for substandard, bulk-generated content that is typically illogical or surreal.
Our research shows that a lot of users appreciate content produced via AI tools, from digital art to informative clips, and we intend to give people the power to increase or decrease of that, based on their own likes,” said a spokesperson.
Volume and Proportions
The service disclosed that there are now over a billion videos on its network labeled as artificially created. Collectively, in excess of 100 mn videos are submitted to the app daily, making AI-generated material a limited part of its complete catalog.
User Control Features
Members will be in a position to reduce or increase the level of artificial media they view by accessing the “manage topic” setting in the app and selecting AI-made media. Further subjects that can already be controlled under the setting include current affairs, clothing, aesthetics, and dance.
Content Labeling and Policies
The platform’s rules require creators to label “realistic” AI-made content, and it prohibits dangerous AI impersonations of well-known individuals or crisis events. Any realistic AI video that is not labeled can be removed under its user guidelines.
Additionally, the platform will now apply an artificially created watermark to content produced via its own AI tools, or flagged by an collaborative effort for recognizing artificial media. This measure is designed to stop users from attempting to evade the labeling process.
Educational Initiatives
TikTok is additionally establishing a $2m awareness program for specialists and institutions to create informative material about ethical AI use. This program will benefit educational bodies specializing in educating computer science.
Content Safety Approach
The platform has defended its content review process, following controversy over intentions to make redundant numerous review team members in Britain. The company is set to cut hundreds of positions in its trust and safety team in its British office, with trade unions and internet security professionals voicing alarm that staff are being displaced by automated systems.
A trust and safety lead clarified that people-led review would remain essential, but automated tools can safeguard staff by filtering out the most distressing content before it has to be seen by people. The company stated a significant reduction in disturbing and violent content reviewed by content reviewers over the previous year due to intervention by its machine learning processes.
“We think that it’s essential to combine people and machines to ensure user safety,” the executive stated. “We’ll have humans as part of that process, but we further consider it’s very important to limit contact with dangerous material as quickly as possible, which is the area in which a extensive automated help is making a difference operate efficiently.”
