• Elon Musk’s generative AI Grok is going open source as soon as this week.
• The AI is part of xAI, a “truth-seeking” AI company founded to compete with OpenAI by Musk.
• Grok told us that it was not aware that it was going open source, but that if it did it could mean more features for X users.

The generative AI platform founded by Elon Musk, xAI, will soon go open-source with its people-facing chatbot, Grok. It will happen as soon as this week according to a post on X from the site’s billionaire owner.

As a chatbot, Grok competes with the likes of Google Gemini, Microsoft Copilot and the current leader in the space OpenAI’s ChatGPT.

Currently, use of Grok is reserved only for paying subscribers of X Premium. Making the AI open-source would mean that users would have access to the building blocks of the Grok generative AI for free. This could potentially lead to Grok being the basis of future generative AI platforms offered by smaller companies or even lone developers.

Google and Microsoft are shying away from open-sourcing their generative AI, while the likes of OpenAI offer the ChatGPT API to paying subscribers. Musk himself has said in the past that generative AI should be freely available to as many people as possible, and warned against big companies using the technology to turn a profit.

More recently, Musk was revealed to have told the OpenAI founders, which he was part of, that the company would not survive unless it pursued profits, in contrast to previous comments, and assertions that “Open” in OpenAI means “open-source.”

He said that the company was founded on a lie.

Grok was specifically launched in 2023 as an alternative to OpenAI’s hold on the market, while Musk said last year that the point of xAI was to “understand the true nature of the universe.”

The chatbot uses a large language model built by former employees of Google’s DeepMind, OpenAI, Google Research, Microsoft and Tesla. Instead of browsing the internet for information to back up its answers, Grok browses conversations on X, formerly known as Twitter.

Unlike chatbots like ChatGPT and Gemini, Grok’s comparison results have not been published, if the AI has even been tested. While GPT-4 has research papers written about it, Grok is more of a fun experiment from the Tesla founder.

From our experiments with Grok, it is evidently less advanced than ChatGPT. For example, it struggles to recall previous prompts asked in the same conversation. We asked Grok today what it going open-source would mean for itself and users on X.

It scanned conversations on the social media site and then told us that it was not aware that it was going open-source.

“My developers haven’t shared any plans with me going open source,” it said, despite thousands of posts on X discussing how Musk plans for Grok to go open source, and posts from Musk himself on the matter.

“If that were to happen,” it said, “it could potentially mean more transparency and collaboration in the development of my AI capabilities, which could lead to improvements and new features for X users. But for now, this is just speculation.”

According to experts who spoke to Reuters, fears about open-sourcing generative AI models like Grok could lead to terrible abuses in the technology. Like terrorists could use the AI to create chemical weapons, or develop conscious super-intelligence beyond human control.

The latter warning seems a bit far-fetched, but we have seen Grok tell people how to make cocaine in the past. Alternatively, the advanced technology could be used to create malware or for hacking. Something that is becoming more prevalent as generative AI advances.