The Clear Line Between Real And AI-Generated Videos Could Vanish In 2025

AI-generated content is here to stay. It’s getting better and better and we’ve seen a significant improvement in AI technologies in 2024. Will we be able to distinguish between what is real and what is not in 2025?

I recently found myself asking, “Is that a real woman?”—while squinting slightly—after seeing a post on Instagram. “No, that has to be AI-generated,” I concluded after a moment. Then I immediately wondered: Will we still be able to tell the difference within seconds next year? All signs point to no, especially on social media, as major recent developments in AI technology suggest.

A few days ago, Intagram’s head, Adam Mosseri, shared a message addressing this issue on Threads. “Generative AI is clearly producing content that is difficult to discern from recordings of reality, and improving rapidly,” he wrote. Mosseri publicly acknowledged that they are having a hard time at Meta tagging AI-created content and controlling and monitoring the massive amounts of media shared every second on their social media platforms.

Mosseri washed Meta’s hands of responsibility and warned users that it’s up to them to discriminate whether a post was real or not. “It’s going to be increasingly critical that the viewer, or reader, brings a discerning mind when they consume content purporting to be an account or a recording of reality,” he added.

Just a few days later, OpenAI released Sora Turbo, its powerful AI video generator, allowing Pro and Plus users to generate realistic videos from text prompts. In just a few minutes users started to create fascinating realistic videos and began sharing them on social media.

Right when it all seemed to reach a new level of video production, Google announced its latest AI video tool, Veo 2, with higher resolution and longer clips than Sora. Of course, the results shared by those who had access to the advanced technology were even more impressive.

“No one can tell this is AI now,” wrote one user.

It’s over.

Google Veo 2 just wiped out the competition and became the new king of the AI media world.

No one can tell this is AI now.

10 wild examples: pic.twitter.com/MKAN4Lwj6m

— el.cine (@EHuanglu) December 17, 2024

It’s Already Happening

During 2024, we’ve witnessed how AI-generated content has impacted populations, elections, and social media users.

In January, New Hampshire residents got a call “from Joe Biden” telling them not to vote in the primary elections. Political consultant Steve Kramer was behind the scam and was fined $6 million for the AI deepfake robocalls. Governments have been understanding the impact AI can have and nations such as Singapore have been developing laws and strategies to help the population flag deepfake content.

In Venezuela, AI-generated videos have been spreading misinformation—but also helping journalists inform through AI avatars to bypass censorship.

People have been having a hard time telling what is real and what is not, and it’s not their fault: AI-generated content keeps getting better and better.

High-quality AI-generated content is a growing business in all its forms; text, audio, and video. Many OnlyFans creators are currently relying on chatbots to engage in conversations with their fans, a job that used to be assigned to gig workers in lower-income countries. The new AI models specialized in flirting techniques have even generated $1,000 in tips from their followers.

In April, Nvidia-backed startup Synthesia launched AI avatars that can express human emotions and speak in 130 languages. TikTok is allowing companies to create AI-generated ads using Getty Images content, and now YouTube creators can use the new AI auto-dubbing feature to translate their content into 9 languages and future updates will replicate the original voice.

Zoom is working on creating AI avatars that look just like users to help them record messages and even participate in meetings. This tool will roll out next year, but people have already been testing multiple tools and companies can already create photo realistic AI clones.

Are you sure you talked to a real person during your last Zoom call?

How To Tell If It Was Created By AI? For Now

We are living in confusing times in this digital era. Filters can make anyone look like a model with flawless skin, and Photoshop-worthy corrections are just a click away on Instagram or Snapchat—how many people would want to share their “natural” image, imperfections and all, when it can be enhanced in seconds? We’re seeing more and more ads featuring photorealistic images generated by artificial intelligence, gradually convincing our eyes of their familiarity and “normality.” And every day, surprising events around the world make us question: what is real, and what isn’t anymore?

The big AI companies have been working on watermarks to spot their AI-generated content and recognize it easily. Google DeepMind launched its open-source watermark system in October to tag AI videos, text, and audio, and Meta also included one in its AI video tool.

But its effectiveness is questionable, starting with Meta just acknowledging that they can’t tag all AI-generated content. Malicious actors can still find ways to remove the watermarks and more laws, partnerships, and agreements are necessary between social media companies and governments to achieve a more successful system.

For now, besides trusting our guts, we can ask ourselves a few questions before sharing on social media or believing what we see:

• Just like Mosseri recommended, consider the following: Who is sharing this information? Is this a trustworthy source?
• Does the person’s expression or movement feel robotic? While AI avatars can be pretty realistic, it is still possible to recognize a weird stiffness or lack of emotion—although consider that a few companies like Synthesia are already improving this.
• Is the content designed to provoke a strong emotional reaction? Malicious actors know how easily we share negative news and interact with emotionally charged material on social media, and research confirms it. Before sharing that content that made your blood boil to get others as outraged as you are, think about whether it’s designed to stir controversy or if it’s actually real news or a real event.
• Can you corroborate the content with other sources? Double-checking with trustworthy sources never hurts.

And finally, have you asked yourself—or your preferred chatbot—how to improve your critical thinking? Developing a critical mind should definitely be among everyone’s New Year’s resolutions.