Pornhub is using machine learning to automatically tag its 5 million videos

The company is starting with facial recognition, but wants its AI to recognize specific categories too

Artificial intelligence has proven to be a dab hand at recognizing what’s going on in photos and videos, but the datasets it’s usually trained on are pretty genteel. Not so for Pornhub, which announced today that it’s using machine learning to automatically catalog its videos.

The site is starting small, deploying facial recognition software that will detect 10,000 individual porn stars and tag them in footage. (Usually this information is provided by uploaders and viewers, who will still play a part by verifying the software’s choices.) It plans to scan all 5 million of its videos “within the next year,” and then move onto more complicated territory: using the software to identify the specific categories videos belong to, like “public” and “blonde.”

pornhub-is-using-machine-learning-to-automatically-tag-its-5-million-videos photo 1

Pornhub’s machine learning models will improve over time as users verify its guesses. Image: Pornhub

In a press statement, Pornhub VP Corey Price said the company was joining the trend of firms using AI to “expedite antiquated processes.” However, the speed at which PornHub’s AI processes the data doesn’t seem like it would be an improvement on its current crowdsourced system. While in beta the machine learning software apparently scanned some 50,000 videos in a month. At this rate it would take nearly a decade to scan the entire site, but presumably improvements are being made.

Price told VentureBeat: “Ultimately, we want to provide our fans with whatever it is they have a penchant for, and our new model will be able to surface more accurate results for them, which will hopefully keep them coming back for. The easier it is to navigate and access what they want, the happier they will be.”

The deployment of facial recognition technology on a porn site has worrying implications for privacy. Revenge porn is often uploaded to popular porn sites, and there may be performers who wish to stay anonymous while publishing videos. Pornhub claims there will be no privacy issues as the system only identifies faces of known performers stored in its database. However, the application of this tech to pornography will still be worrying, especially considering past reports of facial recognition being used to identify porn stars’ real names.

    Share on
    Article Pornhub is using machine learning to automatically tag its 5 million videos compiled by www.theverge.com