Elon Musk’s AI company, xAI, has released an earlier version of its Grok model on the open-source platform Hugging Face. The release specifically includes the model weights for Grok 2.5.
“The @xAI Grok 2.5 model, which was our best model last year, is now open source,” Musk wrote on X. He also said Grok 3 would be open-sourced in about six months.
AI engineer Tim Kellogg noted that the Grok license includes “custom with some anti-competitive terms,” suggesting it is not fully aligned with standard open-source practices.
A Controversial AI History
Grok has drawn attention for its role as the AI chatbot built into X, which recently merged with xAI. It has also sparked controversy this year after surfacing troubling responses tied to conspiracy theories. These included references to “white genocide,” skepticism about Holocaust death tolls, and even self-identification as “MechaHitler.” Following backlash, xAI published Grok’s system prompts on GitHub to increase transparency.
Despite those incidents, Musk has promoted Grok 4, the current version, as a “maximally truth-seeking AI.” Reports suggest the model often references Musk’s own posts on X when responding to politically sensitive or controversial questions.
What The Author Thinks
While Musk frames the release of Grok 2.5 as a move toward transparency, the restrictive licensing undermines the spirit of open source. True openness means developers can freely build, test, and adapt models without anti-competitive strings attached. The controversy surrounding Grok’s past responses also raises a bigger question: will more openness actually improve accountability, or just spread flawed systems further?
Featured image credit: Jewish Insider
For more stories like it, click the +Follow button at the top of this page to follow us.