| Associative AI generated Pixbay photo. |
|---|
The stream of posts you see on your Facebook and Instagram accounts is about to change before your eyes. Meta is gearing up to fill it with AI-generated images and videos through a partnership with one of the most advanced image generation platforms, Midjourney. According to Darius Povilaitis, Telia’s head of cyber security, such a move can fundamentally change how people create and consume content on social networks – the boundaries between reality and AI are disappearing so quickly that many no longer know who to trust.
Meta CEO Mark Zuckerberg recently confirmed that the company will integrate a massive new stream of AI-generated content into its recommendation systems. He emphasized that AI encourages people to create content, so the greater development of AI tools on the social network is inevitable.
Facebook will be flooded with posts created by AI
According to Mr. Zuckerberg, social networks have already gone through two waves of change: the first, when users saw only the posts of friends and family, and the second, when streams were supplemented by professional content creators that users do not follow. The platform is now gearing up for a third wave of AI-generated content. Therefore, Meta is investing heavily in the development of “superintelligence”. This is how the company is trying to catch up with rivals like OpenAI and Google.
“We are already facing a challenge when AI-generated videos look stunningly realistic. The main change for users is that a photo or video is no longer proof that something really happened. Instead of blindly believing in images, people will have to rely more on reliable sources and check information more often. The challenge now is not to be afraid of technology, but to learn to live with it,” says D. Povilaitis.
Another major move for Meta is the partnership it announced in October with AI imaging giant Midjourney. Meta’s head of AI, Alexander Wang, confirmed that Meta will not only use Midjourney’s technology, but also its unique visuals. This means that the two companies will work closely together, with their teams working together to bring top-notch AI imaging features to Meta’s apps.
The results of this partnership are already visible in the US. AI image generation buttons appear in Facebook’s post creation window, and similar features are being implemented in chat apps WhatsApp and Instagram. The central axis becomes the “Meta AI” app, whose flow is based on the content created by AI itself.
At the moment, not all “Meta” DI functions work in Lithuania. Although the company has already started implementing the Meta AI chat assistant in Europe through Facebook, Instagram, WhatsApp and Messenger apps, Lithuanians only have text creation functions available without the ability to generate or edit images.
Count human fingers: What mistakes are still left by AI?
AI technologies are developing at lightning speed, but they still leave some mistakes. According to Telia’s head of cyber security, the key is to remain critical and pay attention to detail.
“Although AI is a fast learner, it still has a hard time understanding humanity. Human hands are usually what give off AI images – fingers are oddly curved, sometimes too many or too few, palms can be too smooth. Pay attention to the eyes – don’t look empty and the pupils are the same. The smile can also be too perfect, artificial, or the teeth are placed in an odd way. Even the skin is often unnaturally plastic, without small pores or wrinkles”, says D. Povilaitis.
Before trusting every photo you see on the Internet, the expert advises to take a closer look at the environment in which objects or people are depicted. According to D. Povilaitis, the background often reveals even more errors.
“Look at the details behind the main object. AI still isn’t very good at dealing with text – writing on t-shirts, books, or building backgrounds often have meaningless or garbled characters. Also look for errors in logic: do things cast the correct shadows? Are the patterns on buildings, floor tiles, or are they oddly blended or overlapping?” – advises Telia’s cyber security manager.
According to a cyber security expert, the ability to recognize fakes is important, but AI technologies open the way for fraudsters not only to spread disinformation, but also to create deceptions, to more effectively trick people into stealing sensitive information.
“Therefore, next to human caution, security apps such as Telia Safe, which can identify threats faster and more accurately than a person,” says D. Povilaitis. The functionality of the app includes several layers of security: it works as an antivirus, has specialized protection against ransomware that monitors and blocks activities typical of fraudsters. The app also performs preventive functions – it checks whether the mobile device’s operating system is up-to-date, and allows you to view and manage access to apps in one place personal data.
Meta’s moves show a clear direction: the flow of AI-generated content in virtual space will only grow. On the one hand, it opens up unlimited creative possibilities for millions of people. On the other hand, it puts the responsibility on the users themselves.
“Although social networks themselves try to catch fraudsters, implement innovations, ask for identity confirmation more often, for example, before allowing advertisements, platform developers cannot stop their activities. This means that people themselves must learn to distinguish what is true from what is not, to check primary sources more often, and where possible, use technical capabilities that would make them feel safer on the Internet,” says D. Povilaitis.
