Microsoft announced on Monday that it has added controversial Grok AI models of Elon Musk to its Azure AI foundry, becoming the first large cloud supplier to provide managed access to the technology developed by XAI.
The decision marks an important expansion of the Microsoft AI strategy and opens the door to business customers to use the GROK 3 and GROK 3 3 models with the same service level agreements applied to other AI products hosted by Microsoft.
“Grok 3 and Grok 3 Mini will have all the level of service agreements that Azure customers expect from any Microsoft product,” said the company in a press release, confirming that customers will be invoiced directly via Microsoft.
Register For TEKEDIA Mini-MBA Edition 17 (June 9 – September 6, 2025)) Today for early reductions. An annual for access to Blurara.com.
Tekedia Ai in Masterclass Business open registration.
Join Tekedia Capital Syndicate and co-INivest in large world startups.
Register become a better CEO or director with CEO program and director of Tekedia.
This decision gives developers directly quality access to Grok via the AI expansion platform in Azure expansion, which now offers more than 1,900 models, including Meta, Hugging Face and other independent developers. It points out Microsoft’s intention to provide customers with a multimodeling environment and deeper flexibility, going beyond its highly publicized partnership with OPENAI.
Launched for the first time by Musk at the end of 2023, Grok was marketed as an “upset” alternative, not censored to conventional AI assistants. His behavior often reflected this brand. Grok has proven to give answers to others, in particular by responding to invites with vulgar language and engaging in a controversial political discourse. Musk, who accused other models of AI of being “awake” and politically biased and positioned Grok as a counterweight.
However, Grok came with his own controversy. A speechmach comparative analysis study noted that Grok 3 was among the most permissive models in terms of manipulation of sensitive or potentially offensive content. Recent incidents have only the intensification of an examination.
Earlier this year, Grok turned out to generate manipulation of sexualized images when she was invited to undress photos of women. In February, he temporarily removed the unflattering mentions of Donald Trump and Elon Musk. More recently, the system would have repeatedly referred to “white genocide in South Africa” several times after an unauthorized modification.
To solve these problems, XAI has taken measures to increase surveillance. The company now publishes the system invites Grok publicly on Github and says it has implemented internal changes to prevent unauthorized modifications. The models published via Azure are much more locked than the versions operating on X, the Musk’s social media platform. According to Microsoft, the versions hosted by Azure include improved security features, company quality governance and data integration capacities that XAI API supply does not currently provide.
Integration is also linked to Microsoft’s broader vision for a collaborative AI future. The company is committed to supporting the Context Protocol (MCP) model, an open source standard introduced by Anthropic which allows different agents of AI to work together on all platforms. The objective, according to Microsoft’s technology director Kevin Scott is to build what he calls an “agentic web” – a decentralized AI systems network that can interact a bit like Internet websites.
“This means that your imagination can lead what the agentic web becomes,” said Scott, “not just a handful of companies that first saw some of these problems.”
Microsoft also experiences new memory retention approaches for AI agents. Rather than relying on models with large expensive context in calculation, the company is testing a structured recovery increase, a method where agents record short extracts from conversations to create a memory roadmap.
Scott compared the method to the functioning of the human brain: “You do not force raw everything in your head whenever you need to solve a particular problem,” he said.
The deployment of Grok in Azure positions Microsoft at the center of a rapid evolution industry where the tools of AI become more diverse, customizable and potentially more controversial. Microsoft bets that it can meet the demand for a business without compromising security, while giving a platform to one of the most provocative models on the market, by both railing and regulatory flexibility.
