Getting your Trinity Audio player ready...
|
Microsoft has launched a programming tool that acts as a semi-autonomous coder and said it would offer AI models made by competitors to OpenAI at its Build developer conference in Seattle.
Microsoft’s new GitHub Copilot programming agent competes directly with a similar tool called Codex announced by OpenAI last week, in a move that shows how Microsoft and OpenAI are now operating at arm’s length, even as Microsoft continues to heavily back the ChatGPT maker.
The GitHub Copilot coding agent is powered by OpenAI competitor Anthropic’s Claude 3.7 Sonnet AI model, Microsoft said.

Semi-autonomous programming
The company said it would offer xAI’s Grok 3 and Grok 3 mini and models from France’s Mistral and German start-up Black Forest Labs, in a further diversification away from OpenAI.
The models are to run on Microsoft’s own Azure data centres, allowing it to ensure availability at a time when resource-hungry AI tools can suffer outages in peak periods.
Microsoft now offers some 1,900 AI models on Azure.
The GitHub Copilot agent can take on programming work in a similar manner to a human programmer, receiving instructions in human language and showing its work for review when it is finished, Microsoft said.
The human developer can then request modifications and when they are satisfied the agent can add the source code to existing repositories.
“The agent excels at low-to-medium complexity tasks in well-tested codebases, from adding features and fixing bugs to extending tests, refactoring code, and improving documentation,” said GitHub chief executive Thomas Dohmke in a blog post.
Previously GitHub Copilot could generate bits of code, but the new agent acts with far greater autonomy, Microsoft said.
The company said at Build that it sees businesses creating their own agents for various tasks inside a business through Microsoft’s Azure Foundry, which allows companies to use a mix of AI models.
Changing relationship
Earlier this month the Financial Times reported that OpenAI and Microsoft were negotiating changes to their partnership, worth more than $13 billion (£9.8bn), to allow the AI start-up to launch a future initial public offering while protecting Microsoft’s access to OpenAI’s future technology.
The companies are also revising terms of a wider contract agreed when Microsoft first invested $1bn into OpenAI in 2019, according to the report.
That contract, which extends to 2030, covers what access Microsoft has to OpenAI’s intellectual property, including models and products, as well as a revenue share from product sales.
Microsoft is reportedly offering to give up some of its equity stake in OpenAI’s new for-profit business in exchange for accessing new technology developed past 2030.