SAN FRANCISCO — Meta has released a new artificial intelligence model that it says rivals technology from OpenAI and Google, and made it available for free for anyone to use.
The new model, called Llama 3.1, extends Meta’s strategy of open-sourcing its models, allowing anyone to use and modify them without paying the company a fee.
If Meta is successful, it could undermine the business models of rival tech giants and make it easier for startups to compete with the likes of OpenAI, while also giving fraudsters, state-sponsored hackers, and other bad actors access to cutting-edge technology.
Meta released the latest version of its AI just three months ago, Llama 3. But the new release includes models trained on much more data than previous versions, potentially improving its capabilities and providing new tools for companies and organizations that want to use larger, more powerful AI models in their products.
“Rama 3 is capable of competing with cutting-edge models,” Meta CEO Mark Zuckerberg said in an open letter on Tuesday. “Beginning next year, we expect future Rama models to be at the forefront of the industry.”
When OpenAI released ChatGPT in late 2022, it set off an arms race among big tech companies to develop new AI products and get people to pay for them. Microsoft signed a multibillion-dollar deal with OpenAI to access its technology, while Google built its own AI models and integrated them into its products. Meta is also spending heavily on AI, but unlike Microsoft and Google, it doesn’t have a large cloud software business that can help it sell its AI to other companies.
Get caught up in
Stories to keep you up to date
Instead, the company has chosen to open-source its AI, hoping that other companies without their own AI technology will use Meta’s technology, creating an ecosystem that will give the company influence across the tech industry, similar to how Google controls the Android operating system and influences the mobile industry.
Meta’s Llama family of AI models has already been downloaded 300 million times by companies and individuals, said Rob Sherman, Meta’s vice president of policy and acting chief privacy officer.
The company’s open-source approach has raised concerns from some politicians, activists and AI researchers that its technology could be used by America’s geopolitical rivals, as well as criminals and fraudsters. Other open-source AI tools have already been used to create child sexual abuse images. But the company has fiercely defended its approach over the past year, and in Zuckerberg’s letter on Tuesday stressed that its open tools are more susceptible to scrutiny by researchers and regulators than the closed systems built by its rivals.
“Open source makes the benefits and opportunities of AI available to more people around the world, de-concentrating power in the hands of a few corporations, and ensuring that the technology is deployed more evenly and safely across society,” he wrote. Meta also provides tools that companies can use to test the safety of their AI systems.
Zuckerberg likened the closed AI model to Apple’s practice of setting rules and charging developers who want to distribute apps on the iPhone, something Meta has had to deal with for years.
“Between the way they tax developers, the arbitrary rules they enforce, and all the product innovation they block from shipping, it’s clear that without rules for phone manufacturers, Meta and many other companies would be able to build much better services for people,” he wrote.
Tuesday’s announcement comes as Meta seeks to chart a new future by building a suite of AI products that it says will change the way people shop and communicate online.
Earlier this year, Meta began integrating Meta AI across its social media apps, allowing it to generate images and answer questions from users in the search boxes of WhatsApp, Instagram, Facebook and Messenger.
But across the tech industry, consumer adoption of AI tools in everyday life remains questionable, as many high-profile AI deployments, such as Google’s integration of AI into its search results, have been beset by missteps and failures that have forced companies to withdraw their products.
“We’re at the stage now where our primary goal is to get hundreds of millions, maybe even billions, of people to use Meta AI as a core part of how they do their jobs,” Zuckerberg told investors in April. “The next step is to build something incredibly valuable.”
