The new Meta.ai website can instantly create stunning AI art

Drafting AI art on Facebook might seem a bit off, but Meta at least adds a watermark to the images.
Meta-AI-burger-mall

Meta has launched Meta.ai, a website specifically geared towards artificial intelligence, featuring a cool functionality not offered by its competitors: it can generate images in real-time, and even animate them as needed.

However, there’s a catch: Meta does want to continue improving Meta.ai, and to do so, it only offers image generation when you log in with your Facebook account.

Meta joins other LLMs or AI chatbots such as Google Gemini, Microsoft’s various flavors of Copilot, Anthropic’s Claude AI (used in Discord), and other sites providing AI solutions. Meta.ai feels similar, but with some limitations: it can’t accept uploaded documents, but it can summarize websites or web pages. Of course, it also has creative purposes: it can also be used to write or rewrite text, much like many other services.

Technically, Meta is upgrading Meta AI and rolling out the website itself. Meta CEO Mark Zuckerberg announced the addition of the Meta.ai website on Threads (also owned by Meta) and an upgrade to the LLaMA 3 model. Meta launched LLaMA in February 2023 and released the 65-billion-parameter model to the public. Zuckerberg stated that LLaMA 3 will also be open source.

But there’s more: “We’re integrating Meta AI into the search boxes at the top of WhatsApp, Instagram, Facebook, and Messenger, making it easier to use,” Zuckerberg added. In other words, Meta’s AI will permeate your social media.

The incredible power of Meta.ai’s AI art

The coolest thing about Meta.ai is its image generation — if you don’t want to tell Facebook what you’re making images of, or generally don’t subscribe to Facebook or Instagram, that’s also its biggest issue. But the progression of images could be unlike anything you’ve seen before. If you’re familiar with how Google or Bing start generating suggestions as you type, Meta does almost the same thing — except it draws actual pictures as you type.

Here is the screenshot of Meta.ai with input prompts. Upon completion, four images will be displayed.


Even though the final images may not be as crisp as some final images, it completes the task at an astonishing speed. After completion, Meta generates four images at 1280×1280 pixels, which can be highlighted or displayed at larger sizes. They are all downloadable and come with an “AI Imagination” watermark. After completion, the images can also be edited to make changes to the finished product.

However, another feature offered by Meta is the ability to animate each generated image. Unlike Runway or Sora; it’s an animation. Instead, Meta.ai generates several frames of video before and after the original image, like a GIF. On my screen, this process took at least 30 seconds to a minute. (On the second attempt, it seemed to freeze and did not complete after several minutes.)

I requested Meta.ai to animate the image of the parrot dressed as a black private detective.


One can’t help but wonder if Meta, as the owner of Facebook, will allow AI-generated art to be created on Facebook as a way to generate erroneous information. Quick attempts to draw Donald Trump resulted in a non-specific error message. However, any images you create can still be downloaded, and the watermark will be cropped out.

Meta.ai has enough functionality that we recommend giving it a try. But I think it’s fair to say that the long-term impact won’t be on Meta.ai itself but on other assets owned by Facebook.

Leave a Reply

Your email address will not be published. Required fields are marked *