Social media giant Meta has added its newest AI assistant to almost all its apps, co-founder and CEO Mark Zuckerberg announced on Thursday.

“We are releasing the new version of Meta AI, our assistant that you can ask any question across our apps and glasses,” Zuckerberg said in a video posted to Instagram. “Our goal is to build the world's leading AI and make it available to everyone.”

The news comes as Meta released the core components of Llama 3 under an open-source license, allowing public use and review.

After its Metaverse ambitions fizzled in late 2022, Meta shifted focus and dove hard into generative AI. Rumors began to swell that Meta would release its Llama 3 generative AI model in May. The company hit publish early on Thursday.


“The bottom line is that we believe that Meta AI is now the most intelligent AI assistant that you can freely use,” Zuckerberg said.

Meta AI uses real-time data from Google and Bing in its responses, Zuckerberg said.

Beginning today, users logging into their Meta apps notice a new, free-to-use Meta AI search bar added to WhatsApp, Instagram, Facebook, and Messenger. Threads, Meta’s Twitter competitor, was notably absent from the announcement. In addition to its social media applications, Meta AI is available through the Meta AI website to logged-in Facebook users.

Zuckerberg said both the desktop and mobile versions can create high-quality images. Once created, images can also be animated into short three-second clips.


The desktop version of the Llama 3 image generator begins composing images while the user is entering a prompt, giving them an instant preview of what the final picture might eventually look like.

Although his video announcement was aimed at the mainstream app user, Zuckerberg did address its decision to open-source the Llama 3 AI model, saying it was the best way to create better and safer products.

“We're investing massively to build a leading AI, and open-sourcing our models responsibly is an important part of our approach,” Zuckerberg said. “The tech industry has shown over and over that open-source leads to better safer, and more secure products, faster innovation, and a healthier market. “

AI-generated deepfakes are an ongoing and escalating problem that’s only exacerbated by the relentless cycle of new and updated generative AI models hitting the market.

In February, the Biden Administration said it would use watermarking and cryptography to tag authentic content and fight political disinformation.

Aiming to get in front of Llama 3 being used to create deepfakes, images created using the tool include an “Imagined with AI” disclaimer at the bottom of the picture. Meta has meanwhile pointed out that its AI development team included guardrails to detect prompts that go against the company’s policies, like asking how to commit crimes.

Meta did not immediately respond to a request for comment from Decrypt.


Edited by Ryan Ozawa.

Generally Intelligent Newsletter

A weekly AI journey narrated by Gen, a generative AI model.