The internet is filled with digital portraits made using artificial intelligence that look vivid, imaginative and enticing, and are everywhere on the internet. As a result of these technologies, experts consider them to be potential tools for aggregating user data.
A plethora of fantastic portraits have started to appear on social media, many of which have been created by Lensa AI, a photo editing app. Lensa AI has recently released a feature called 'magic avatars'. At the beginning of December, the iOS App Store ranked it at the top of the "Photo & Video" category owing to its popularity.
To generate a portrait of a person, the app requires at least 10 pictures of the person. Once the system has calculated the numerical value for each facial point, it adds these values all up. This creates a digital image of the person by adding up the numbers. Even so, there are consequences to the facial scan that is required.
According to Juergen Schmidhuber, an internationally recognized computer scientist and leader in artificial intelligence, companies want to get you to give them your data. In exchange, you get something back, which is a pleasurable experience. "At the moment, it is just about selling faces and ads," he continued. "But it will be much more dangerous."
Data that pertains to your face
Before Lensa can create a portrait for you, you have to pay $7.99 and submit pictures. In addition, you have to agree to Lensa's terms and conditions in order to use their service.
Lensa's privacy policies state that they collect, store, and use your Face Data for online processing purposes. These Face Data will be automatically disposed of after 24 hours of processing. As far as the Magic Avatar feature is concerned, the photos that are taken by the Magic Avatar feature will be removed when they have been generated by AI.
One of the reports states that Mari Galloway, a cybersecurity expert familiar with artificial intelligence and consumer technology, believes users should remain cautious about Lensa's privacy policy, regardless of what it says.
As she states, "We don't know what they intend to do with that data, what they intend to do with that information." "When it comes to photos and videos, they do not keep them longer than 24 hours." The question is, do we actually know what they are doing with it?
What is the process that they use to delete it? What kind of encryption is used to protect the data? In this area, we do not really know what they really mean since they haven't really given us any information on that topic."
As Schmidhuber explains, the simple act of deleting a post on Facebook can prove to be extremely difficult. He asks how Lensa removes such large amounts of data so easily and regularly.
A tweet from Prisma Labs, the company behind Lensa, on Dec. 6 stated that as soon as the avatars are generated, the user's photos are permanently deleted from Lensa's servers, reiterating the aforementioned claim in the company's privacy policy that the user's photos are permanently deleted by the company.
A second concern that Galloway raises is whether the company will continue to operate if its ownership changes?
Even if there is still data in their database, they will pass this information on to the next person, says Galloway.
Galloway also added that it's "really imperative" that we safeguard our personal information in order to ensure it does not end up in the wrong hands someday.
In order to protect yourself, you need to be very careful about what you put on your phones and what you share with these apps, she says in a warning to consumers.
All kinds of propaganda purposes
As Schmidhuber explains, artificial intelligence images can only be created by using images of real people, in many cases. Artificial intelligence gets smarter and smarter as there are more image databases that are available. The company can use this data to generate additional faces for further training of networks in order to be at least as good as the original ones, according to him.
Whenever the database has a sufficient number of images, it will be able to create images based on those images, as Schmidhuber stated. When technology progresses to the point that it comes to making digital people who can speak and appear like real people, the dangers of AI become evident.
In addition, he said that you could use the application for all kinds of inappropriate purposes, such as when it is not needed. In order to generate a sequence of frames, the neural network has to take a lot of time. This might make it look like a video showing this person talking, but may, in truth, show something this person has never said.
