Sam Altman explains OpenAI’s shift from open to closed AI models

Sam Altman explained OpenAI’s thinking on open- and closed-source models in a recent Reddit AMA.

Why doesn’t OpenAI open-source the AI models behind ChatGPT if the company is called OpenAI?

It boils down to what OpenAI is “good at” and the pathway to hitting the company’s safety thresholds for its AI models, CEO Sam Altman said in a Reddit Ask Me Anything session Thursday.

“OpenAI has shifted from a more open-source approach to a more closed model in recent years,” a Reddit user said. “Can you elaborate on the reasoning behind this change, and how you weigh the trade-offs between openness and the potential risks associated with widely accessible advanced AI technologies?”

Altman responded that open source “plays an important role in the ecosystem and there are great open source models in the world.”

“We also think there’s an important role in the world for powerful and easy-to-use APIs and services, and given what we are good at, we see an easier way to hit the safety threshold we want to hit this way,” he said. “We are pretty proud of how much value people get out of our services.”

Altman added that he “would like us to open source more stuff in the future.”

The comments from Altman follow a year in which OpenAI has received criticism for no longer open-sourcing its AI models. The company previously released the code for its GPT-2 AI model in November 2019. However, it did not release the code for its future models, including its most recent GPT-4o and o1-preview models.

Elon Musk, who cofounded OpenAI in 2015 alongside Altman and others before leaving the company’s board, has been one of the most vocal critics of its new approach.

In December 2022, a month after ChatGPT came out, he tweeted, “OpenAI was started as open-source & non-profit. Neither are still true.” OpenAI became “capped profit,” a hybrid of nonprofit and for-profit structures, in 2019.

Musk filed a lawsuit against Altman and OpenAI in March, claiming a breach of contract and that the company has abandoned its original mission. OpenAI published a blog post responding to his claims shortly after.

One email exchange published in the blog post showed Ilya Sutskever, one of OpenAI’s cofounders and its former chief scientist, writing to Musk, “As we get closer to building AI, it will make sense to start being less open. The Open in openAI means that everyone should benefit from the fruits of AI after its built, but it’s totally OK to not share the science (even though sharing everything is definitely the right strategy in the short and possibly medium term for recruitment purposes).”

The screenshot OpenAI published showed Musk responding simply, “Yup.”

Musk dropped the lawsuit in June before filing a new lawsuit in August alleging he was “deceived” into cofounding the AI firm.

Whether to be open- or closed-source is a hot topic in the AI industry right now. One of OpenAI’s competitors that touts its open-source efforts is Meta, whose latest offering in the space, the Llama 3.2 models, came out in September. CEO Mark Zuckerberg has said he believes open-source is “safer than the alternatives” and “necessary for a positive AI future.”

It’s worth noting the company requires firms with a user base of 700 million active users a month or more to request use of Llama, and Meta can decide whether to grant access or not.

The nonprofit Open Source Initiative has previously said this license “would not be qualified as an open-source license because it contains restrictions on commercial needs.” The OSI also recently said Meta’s Llama doesn’t satisfy its definition of open-source AI.

A Meta spokesperson said the company disagrees with OSI’s definition.

“There is no single open source AI definition, and defining it is a challenge because previous open source definitions do not encompass the complexities of today’s rapidly advancing AI models,” the spokesperson said.

“We make Llama free and openly available, and our license and Acceptable Use Policy help keep people safe by having some restrictions in place,” they added. “We will continue working with OSI and other industry groups to make AI more accessible and free responsibly, regardless of technical definitions.”

Elsewhere in the Reddit AMA, Altman said OpenAI’s GPT-5 model is likely not launching this year and that he was using ChatGPT to respond to some of the online questions.