Stay informed with free updates
Just sign up for artificial intelligence myFT Digest — delivered straight to your inbox.
TikTok becomes the first social media platform to automatically label some content generated by artificial intelligence, as rapid advances in generated AI deepen concerns about the spread of online disinformation and deepfakes.
Online groups such as Facebook owner Meta and TikTok are already asking users to disclose whether realistic images, audio or videos are created through AI software.
The viral video app, owned by China's ByteDance, announced Thursday that it will go one step further and introduce a unique feature that will label videos that can be identified as AI-generated as such. This includes content created using Adobe's Firefly tool, TikTok's own AI image generator, and OpenAI's Dall-E.
“The challenge, as we know from the many experts we work with, is… harmful content generated by AI,” said TikTok's head of operations and trust and safety. said one Adam Presser.
“This is so important to our community because authenticity is one of the things that makes TikTok such a vibrant and fun community.” . They want to be able to understand what is created by humans and what is enhanced or generated by AI. ”
Social media platforms like TikTok, Meta, X, and YouTube are all exploring ways to integrate generative AI into their platforms through chatbots and new tools to help influencers and advertisers create media. However, these platforms have been criticized for allowing users to be bombarded with low-quality spam content generated by AI.
In a year of major elections around the world, these companies are putting in place guardrails against misleading deepfakes, curbing covert influence operations, and ensuring content is properly moderated while remaining nonpartisan. They also face pressure to do so.
Earlier this week, TikTok and its parent company ByteDance filed a lawsuit against the U.S. government, challenging laws aimed at forcing them to sell and ban the app. Lawmakers had expressed concern that the platform could promote disinformation and propaganda.
TikTok announced Thursday that it is joining a coalition of technology and media groups led by Adobe that will incorporate so-called content credentials into AI-generated products.
The technology embeds a digital fingerprint into multimedia AI content, along with other identifying information such as when, where, and by whom the material was produced. TikTok uses these indicators to automatically flag when content is created using AI products.
OpenAI announced Tuesday that it will join the coalition and embed fingerprinting technology into all images produced by its image model Dall-E 3. Eventually, the ChatGPT maker said it will also embed these in its video generation model Sora. Widely released.
Big tech companies like Google, Microsoft, and Sony are considering incorporating this technology into their AI tools.
Meta announced earlier this month that it would start stamping AI-generated content with “Made by AI” labels by detecting invisible markers inserted by groups such as Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock. Announced. The Facebook owner also said that he is developing a deepfake detection classifier.
Experts say malicious actors and sophisticated disinformation groups are likely to use open-source AI-generating tools to create deepfakes, making them difficult to track and vulnerable to digital fingerprinting and flagging. He claims that he will not be able to stand up.
Tech companies say their efforts are a first step toward tackling the problem.
Adobe's General Counsel and Chief Trust Officer Dana Rao said: “The premise of this solution is that if you want to be transparent and have authentic and transparent conversations with the public, this tool allows you to do that in a world where everything is done digitally.” can be operated. ”
