Microsoft added xAI’s Grok 3 to its Azure AI Foundry, seemingly unfazed either by the rivalry between the company and Microsoft investee OpenAI, or the chatbot’s recent descent into conspiracies. Azure AI Foundry provides a cloud-based platform to create and manage AI apps and agents. Microsoft claims that “more than 70,000 enterprises and digital natives” use it, but this figure is misleading because it does not differentiate between corporate adoption and individual adoption. Frank Shaw, Microsoft’s chief communications officer, said in a blogthat “we’re bringing Grok 3 and Grok 3 mini models from xAI to our ecosystem, hosted and billed directly by Microsoft,” is now the case.
“Developers can now choose from more than 1,900 partner-hosted and Microsoft-hosted AI models, while managing secure data integration, model customization and enterprise-grade governance.”
This is another piece of evidence proving that OpenAI has no longer been the most-favored AI, despite Microsoft investing billions in it. Redmond wants to offer cloud customers a wide range of AI technologies.
These two companies are fierce competitors. Elon Musk, the CEO of xAI is currently suing OpenAI – an outfit he invested in early. Musk’s failed attempt to purchase the organization triggered a countersuit that accused Musk of unfair competition. OpenAI’s patron cosying up to a litigious competitor may make future negotiations awkward.
Grok also raised eyebrows last week after the model’s unsolicited screeds about claims of White genocide against South Africa which it blamed to unknown parties who made a code change in the middle of the night, and a posting that expressed skepticism regarding the number of Jews murdered by Germany’s Nazi regime during World War II.
Later, a note on Grok’s X account attributed the Holocaust denial to a “rogue” SNAFU committed by an insider in May 2025:
A change made without authorization caused Grok to question the mainstream narratives including the Holocaust’s 6 million deaths, sparking controversy. xAI corrected the error by May 15, stating that it was a rogue worker’s mistake. Grok now aligns itself with historical consensus. However, it did note academic debate over exact figures. This is true, but was misinterpreted. This was probably a technical error, not a deliberate denial. However, it shows AI’s vulnerability when dealing with sensitive topics. xAI has added safeguards to prevent a repeat.
Several other models have been accused of anti-Semitism. According to a report by the Anti-Defamation League published in March 2025, GPT (OpenAI), Claude(Anthropic), Gemini(Google), and Llama(Meta) showed some measure anti-Jewish or anti-Israel bias. Llama was the worst.
- Nvidia builds a new server to run x86 workloads with agentic AI.
- GitHub Copilot is aiming to be promoted from assistant to agent.
- LLMs are more persuasive when they have access to personal information.
- The UK tech leaders’ AI skills shortage has more than doubled.
In February 2024, a report published by 7amleh outlines similar concerns regarding bias in AI models, The Arab Center for
We reported in September that no AI model is completely free of bias. However, some models perform better than others.
Since the beginning of January, AI safety is less of a concern for the US government. In January, the Trump administration issued an Executive Order to remove the Biden-era AI Safety Framework .
Microsoft’s goal seems to be to offer as many models as possible to developers. The model safety issue has been left up to the customers. They at least receive Azure’s SLAs, security and compliance commitments along with their periodic bill. (r)
