Microsoft to host Elon Musk's Grok AI

SEATTLE: Microsoft has agreed to host Elon Musk's Grok AI chatbot on its cloud servers. 

The deal comes just days after Grok made headlines for giving offensive responses with comments about “white genocide” in South Africa. Despite the controversy, both companies say the move will improve the chatbot and help keep it in check.

Musk told an event hosted by Microsoft that his company’s models “aspire to truth with minimal error,” adding, “There’s always going to be some mistakes that are made.”

The Grok chatbot recently stirred controversy by responding to unrelated user prompts with unfounded right-wing claims about the supposed oppression of white South Africans.

In a recorded conversation with Microsoft CEO Satya Nadella, Musk said xAI would always admit when its Grok AI models made errors.

“It’s incredibly important for AI models to be grounded in reality,” said the Tesla tycoon.

Generative AI models are often pre-programmed by engineers — through what are known as system prompts — to avoid certain answers or maintain a specific tone or style, regardless of what the user inputs.

Recently, the latest model from industry leader OpenAI was found to be producing overly flattering responses, prompting the company to act swiftly and fix what it called a bug.

Grok’s recent replies raised alarm because they echoed a conspiracy theory often shared on social media by Musk, who was born in South Africa.

The company did not name the person behind the code change but said an “unauthorised modification” caused Grok to respond in a way that breached xAI’s internal policies and core values.

In response to the backlash, the startup announced it would make Grok’s system prompts public, overhaul its review processes, and establish a “24/7 monitoring team” to handle future issues.

Although he did not directly mention the incident, Musk told the Microsoft event that xAI would be transparent when errors occurred.

This was seen by some as a subtle jab at OpenAI, creator of ChatGPT, which is Microsoft’s main partner in building its in-house Copilot models.

OpenAI, co-founded by Musk in 2015, is often criticised for keeping its internal technology secret, unlike more open-source efforts such as Meta’s Llama or the models from Chinese company DeepSeek.

OpenAI CEO Sam Altman also took part in the Microsoft Build event in Seattle via a live video call with Nadella, where both leaders highlighted progress in their joint AI ventures.

‘Virtual teammate’

The Grok models from xAI will be hosted on Microsoft’s Azure AI Foundry — a platform that offers hundreds of models for developers to use in building their own generative AI tools.

The platform already includes models from OpenAI, DeepSeek, Mistral, Meta, Stability AI, and now xAI.

Microsoft’s chief executive said AI coding tools are rapidly becoming “agents” that act like team members to help developers.

Some 15 million developers have used GitHub Copilot — Microsoft’s AI tool for coding — to write or debug code, the company said.

“This is one of the biggest changes to programming that I’ve ever seen,” Altman said during his exchange with Nadella.

“This idea that you now have a real, virtual teammate, that you can assign work to.”

Last week, Microsoft announced plans to cut unnecessary layers of management and embrace AI-driven efficiency, as media reports indicated the company was laying off thousands of workers.

While Microsoft did not confirm the total number of job losses, US media reported that around 6,000 positions — about three percent of its global workforce — would be cut.