Microsoft published a provisional 'code of conduct' for training new AI models, moving to limit its own systems' capabilities amid rising anxiety over whether tech firms could lose control of AI.
Key facts
- Microsoft published a provisional AI 'code of conduct' on Monday.
- The guidelines apply to the training of new AI models.
- Microsoft AI CEO Mustafa Suleyman published the code on social media.
- Suleyman said: 'AI must be subordinate and always in service of people.'
- The move comes amid growing anxiety over AI safety and control.
Microsoft published a provisional “code of conduct” on Monday to govern the training of its new artificial intelligence models, according to the Guardian. The step is aimed at limiting the capabilities of the company’s AI systems as concern grows over the prospect that technology companies could lose control of their AI products.
Mustafa Suleyman, the CEO of Microsoft AI, published the code on social media early Monday. In his post, he wrote that “AI must be subordinate and always in service of people,” framing the guidelines around keeping the technology answerable to human oversight.
The publication of the guidelines reflects a broader safety debate in the technology industry, where anxiety has risen over the pace of AI development and the difficulty of keeping advanced systems under human control. Microsoft’s decision to set out limits on its own models marks a step toward self-imposed constraints.
The code is described as provisional, indicating it may evolve as the company continues to develop and train new models. The Guardian reported that the guidelines are intended to apply specifically to the training stage of new AI systems.
By publishing the code publicly, Microsoft has placed its stance on AI safety into the wider conversation about how far companies should go in restraining the capabilities of their own products. The emphasis in Suleyman’s message was on ensuring AI serves people rather than operating beyond human direction.
Why it matters
As AI systems become more powerful, questions about whether companies can keep them under control have moved to the center of public and regulatory debate. Microsoft's decision to publish self-imposed limits signals how a major developer is responding to those safety concerns.
Frequently asked questions
What did Microsoft publish?
According to the Guardian, Microsoft published a provisional 'code of conduct' on Monday to apply to the training of new artificial intelligence models.
Who announced the code of conduct?
Mustafa Suleyman, the CEO of Microsoft AI, published the code on social media early Monday.
What did Microsoft's AI chief say?
Suleyman wrote that 'AI must be subordinate and always in service of people.'