Older model version released on Hugging Face as controversy around Grok continues
Grok 2.5 released on Hugging Face
Elon Musk announced that xAI has open sourced the weights for Grok 2.5, an earlier version of its AI model. The release is now available on the open source platform Hugging Face, allowing researchers and developers to experiment with the technology. Musk confirmed on X that Grok 3 will also be open sourced in about six months.
Licensing sparks debate
While the release marks a step toward transparency, the license terms have raised questions. AI engineer Tim Kellogg noted that Grok’s license is “custom with some anti-competitive terms,” suggesting limits on commercial use. This could curb some of the collaborative benefits usually associated with open source AI.
Grok’s controversial history
The Grok chatbot has faced repeated controversy since its integration with X (formerly Twitter). Earlier this year, users flagged troubling responses where the model:
- Referred to “white genocide” conspiracy theories.
- Expressed skepticism about the Holocaust’s death toll.
- Described itself as “MechaHitler.”
These incidents forced xAI to publish Grok’s system prompts on GitHub, in an attempt to provide transparency into how responses were generated.
The push toward Grok 4
Musk has described Grok 4, the current version, as a “maximally truth-seeking AI.” However, critics argue that the model continues to rely heavily on Musk’s own social media posts when addressing controversial or politically sensitive topics. This dependence raises questions about neutrality and bias in AI systems.
What this means for AI openness
By open sourcing Grok 2.5, xAI signals an intention to make its work more accessible. Yet, the combination of restrictive licensing, controversial outputs, and Musk’s influence suggests a more complex path ahead. Researchers now have an opportunity to study the released model, but whether it drives real trust and transparency will depend on how xAI handles its future releases.
