Meta AI – On the exploitation of artists on Instagram

Since the beginning of June, artists have been fleeing Instagram to protect their work from Meta's AI . Meta announced that it will initially train its AI on artists' creations on Instagram in the US and Latin America – without compensation or an opt-out option. This is not yet the case within the EU.

However, we should also pay attention here in our own country and follow current events closely. To protect themselves, more and more artists are turning to the AI-skeptical app Cara .

What had happened?

Painters, photographers, and other artists have long used Instagram to showcase their work and increase their visibility. But now, many are leaving the platform due to concerns that Meta, Instagram's parent company, is using their art to train AI models, as the Washington Post reported earlier this month.

As a sign of protest, artists are once again sharing messages and templates on other platforms. Many are switching to Cara, an app specifically for artists that prohibits the use of their content for AI purposes.

The outrage arose after a Meta manager announced in May that public Instagram posts would be used as training data. A few weeks later, the company informed users in Europe that their would be used for AI training starting June 26 without the option to opt out .

In the European Union, however, people have the right to decide how their personal data is used. Tensions between online creators and AI companies have increased. Everything publicly published on the internet is currently considered potential training material for AI systems . This could lead to those who created the data—such as authors, musicians, and visual artists— being replaced by AI .

Artists feel powerless – they rely on meta-apps to market themselves, but cannot prevent their work from being used for AI purposes. Some are already on the verge of losing their livelihoods.

What is Meta AI?

Meta AI is an innovative product based on machine learning and artificial intelligence that can be seamlessly integrated into various meta-platforms such as Facebook, WhatsApp, and Instagram. It also offers the option of standalone use.

AI model is based on a variety of sources, including publicly available online information, licensed data, and insights gained from the use of Meta products and services.

Meta officially announced that Meta AI has recently been trained on Facebook and Instagram posts and photos. The statement reads:

Publicly shared posts from Instagram and Facebook – including photos and text – were part of the data used to train the generative AI models that power the features we announced… . We did not train these models on other people's private posts. Nor do we use the content of your private messages with friends and family to train our AI

Meta has published a document titled "How Meta uses information for generative AI models and functions" that explains the details.

Meta's new Privacy guidelines and the powerlessness of users

Meta's latest change to its Privacy policy has disappointed many of its 1.2 billion users worldwide, following the initial revelation last month.

By agreeing to Meta's terms and conditions for one of its products, users in many parts of the world automatically agree that Meta may use their information to develop its AI models.

Photos, posts, original artwork, and other creative resources from artists can now be used by Meta to train the company's AI image generator. This policy takes effect today, June 26, 2024.

As Wired reported, shortly after Meta's announcement, more than 130,000 Instagram users shared messages on Instagram stating that they did not consent to the use of their data to train AI.

However, according to current legislation, creators on Instagram appear to have unwittingly sold "META-physical"

Only in the European Union do strict Privacy guidelines and intellectual property rights still prevent the Meta Group from making these adjustments in this country as well.

Cara as a refuge of protest

Cara's founder, Jingna Zhang , proudly announced that the app had grown from 40,000 to 650,000 users in just one week. At one point, it even ranked among the top five social apps in the Apple Store. Whether Meta will be impressed by this success remains to be seen. "It was an all-nighter ," reported Zhang, who is not only a photographer but also a dedicated advocate for artists' rights. "This exceeds our expectations."

Artists like Zhang have taken legal action against AI companies such as Google and Stability AI , claiming that they train their algorithms with copyrighted material from the internet. Authors and publishers like George R.R. Martin , as well as The New York Times, have also taken similar legal action. The companies argue that the training material falls under "fair use" laws , which permit remixes and interpretations.

Zhang recently mentioned that the free Cara app, launched in January 2023, crashed several times this week due to technical issues, despite its continued popularity. The app's home tab, available on iOS, Android, and the web, resembles an Instagram feed and offers features such as "Like," "Comment," and "Repost.".

Ben Zhao , a computer science professor at the University of Chicago, observed with concern how various platforms make false promises and deceive users. Some platforms for artists are even evolving into so-called “AI farms .” Zhao and his colleague Heather Zheng developed the tool Glaze to protect the originality of artworks from AI imitation and successfully integrated it into Cara.

According to Cara's FAQ, artists are prohibited from publishing AI-generated works until "far-reaching ethical and privacy legal issues" are resolved. The AI ​​company Hive's detection technology (AI Detector)  is used to monitor for violations and tag uploaded images with a "NoAI" tag, thus protecting them from unauthorized copying.

Nevertheless, the risk remains that AI companies could still use the images. Some artists are already reporting on the impact of AI on their business success.

How Generative AI is hurting artists

artist and illustrator Kelly McKernan joined Facebook and Instagram over a decade ago and has successfully used these platforms to attract customers.

However, in 2022 and 2023, she experienced a 30 percent drop in income due to the increasing prevalence of AI-generated images on the internet, as she stated to the Washington Post.

When she Googled her name, she discovered that the first result was an AI-generated image in her style, which she found frustrating. Meta's AI scraping policy ultimately proved decisive for her. Together with two other artists, McKernan filed a lawsuit against AI companies such as Midjourney and Stability AI.

Freelance illustrator Allie Sullberg decided to download the Cara app two weeks ago after noticing that many of her fellow artists on Instagram were discussing AI scraping and switching to Cara. She expressed her frustration that Meta portrays its AI initiatives as beneficial for creators, even though they don't actually derive any significant benefit from having models applied to their work.

Users of meta standards automatically agree to the AI ​​guidelines when they use the platforms in accordance with the Privacy policies and terms. Sullberg stated that she first became active on Instagram The first generative image model for end users, DALL-E by OpenAI , was launched in 2021.

A spokesperson for Meta told the Washington Post that the company does not offer an opt-out option.

However, depending on where people live, they may raise concerns about the use of their personal data for AI development in accordance with local Privacy .”

so Richards.

this is an unparalleled display of disrespect towards the millions of artists on Instagram to whom Meta owes a large part of his success. It feels like a slap in the face.

Jon Lam , a renowned video game artist and copyright advocate, has been searching extensively for solutions to stop AI scraping on Instagram. After discovering a corresponding form, he found that it only applied to users in Europe, where strict Privacy regulations are in place.

Lam stated that he “deeply angry and upset” towards Meta and other AI companies.

These companies have strayed from their origins. We have been given a false impression that social media is for maintaining contact with friends and family and sharing activities.”

so Lam.

A decade later, they are merely platforms on which data can be collected and training data generated.”

McKernan expressed hope that the ongoing lawsuits would put pressure on AI companies to change their policies.

Complacency has allowed companies like Meta to treat make their money

That was her statement.

European Privacy (still) saves artists from Meta's data hunger in the EU

Creative professionals within the European Union can breathe a sigh of relief for now. The attempt by Facebook's parent company Meta to use public user posts in Europe to train its AI models has been temporarily thwarted following pressure from data protection advocates.

Joachim Rodriguez y Romero

Owner and Managing Director of Kunstplaza . Publicist, editor, and passionate blogger in the fields of art, design, and creativity since 2011. Graduated with a degree in web design from university (2008). Further developed creative techniques through courses in freehand drawing, expressive painting, and theatre/acting. Profound knowledge of the art market gained through years of journalistic research and numerous collaborations with key players and institutions in the arts and culture sector.

www.kunstplaza.de