Lensa AI: Tool that turns your photos into stunning portraits hit by growing criticism

(Getty Images)

A tool that generates stunning portraits using AI is being hit by increased criticism over its ethics and privacy.

Lensa AI has rocketed to the top of the App Store, with its artistic portraits becoming a rapid success. It works by having users feed its system 10 or 20 photos of themselves, which are then fed through a model to turn them into portraits in a variety of artistic styles, and for which they must pay to then get access to images.

Those images have gone viral as users post them on social networks and begin to use them as avatars.

But that same success has brought concern over the ways that the AI system and others like it could be violating both the privacy of people using it and the artistic ownership of the people who created the styles from which it borrows.

The concerns about privacy have dogged many photo editing apps, especially those that have seemingly become popular very quickly. (The app has been around since 2018, but the AI portrait photo was added in November and much of its popularity has happened since then.)

In 2019, for instance, similar concerns sprung up around FaceApp, which allows users to upload photos and have them turned old, and then the same thing happened in 2020 with another app called Reface that let people put their image on top of famous GIFs. The images went viral – but then so did worries about how the company could use the images that were uploaded, and whether they were safe.

In both cases, the apps looked to assure users that the images were only being kept so that the app could work, and that those images would not be abused. But the concern sprung from similar wording in their terms and conditions which was fairly general about what rights users had over the images they were uploading.

Lensa AI has been subject to similar concerns, as a result of similar wording. Its conditions give it “all rights in and to your user content”, which includes the option to use them in a wide variety of different ways.

The reason many apps include such conditions is often because writing them in the most expansive terms allows companies to avoid the risk of overstepping their permissions. But it also means that users are left with little way of controlling the use of their images once they are uploaded, and no real form of recourse should they be upset by what is happening to them.

Lensa says in its privacy policy that the images are used only for the effects. And the company told TechCrunch that the images are deleted once they have been used in its AI system, though that too has led to some concern that people’s images are being fed into a dataset that they might not fully understand.

Other worries have been expressed about the content of those images, rather than how they are being produced. The system used to create them – called Stable Diffusion – was trained on a set of images that will inevitably include some biases, and many users report that the system appears best when used with photos of white people, who may have been more present in the original data.

Stable Diffusion has also faced criticism over the way that it is able to be used to create pornographic material, without the consent of the person depicted, and since Lensa AI is built on the same system it could run into some of the same problem. An ongoing debate is happening within the AI art community over whether it should be possible to use the system this way – and a recent update brought changes that made it harder to make NSFW images and more.

Other criticism has focused on such AI systems more generally than just in the case of the newly popular portrait app. Many such systems have been accused of stealing the style – and sometimes actual images – of artists, and then using them without permission.

On some systems, for instance, it is possible to explicitly command the AI to generate a picture in the style of a specific artist. (This was another change in the recent update to the Stable Diffusion model.) The artist is not recompensed for their style being used – and may not even have been aware that their art was fed into the system in the first place – and so critics have argued that such tools are taking valuable money and credit from visual artists.

Prisma Labs, the company behind the app, looked to address some of the concerns about the way it works in a thread of tweets. It addressed both the ethical concerns about what it is doing to art as well as worries about what might be happening to the images that it produces.

“As soon as the avatars are generated, the user’s photos and the associated model are erased permanently from our servers. And the process would start over again for the next request,” it wrote in one post.

In another, it claimed that the use of AI in art would not make artists useless.

“AI produces unique images based on the principles derived from data, but it can’t ideate and imagine things on its own. As cinema didn’t kill theater and accounting software hasn’t eradicated the profession, AI won’t replace artists but can become a great assisting tool,” it wote.

“We also believe that the growing accessibility of AI-powered tools would only make man-made art in its creative excellence more valued and appreciated, since any industrialization brings more value to handcrafted works.”

But that thread led to yet more criticism, in particular from people who noted that the data used to train the model was people’s artistic images, and that in so doing they might be stealing their style.

“You are using others’ work to mimic and train the computer and replicate the artists style.” wrote one user. “I will never support any company like this, and I can see many others feel the same.”