Was Microsoft right to place limits on how far it develops its most advanced AI models?
On September 14, 2026, Microsoft released a code of conduct putting limits on its future AI models, days after OpenAI and Anthropic leaders agreed to slow frontier development amid warnings from AI researchers.
It was a smart move but not for the obvious reason. MS was not going to be one of the winners. It maintains its relationship with OpenAI. MS is a mature business and isn't wired for the sort of risks associated with becoming an AI leader. Do what you do well and build partnerships for what you don't do well.
They (and maybe Meta) have the least to lose if the pace of progress is constrained.
And the least to gain if it isn't. I think MS is playing this intelligently and appropriately conservatively. META in my humble analysis is an octogonal, rotating train wreck that manages to survive only because of the legacy Facebook product and the huge ad revenues it produces. Zuckerberg had one good idea. End of story.
Agree re: Zuckerberg. Not strong on innovation (e.g., $80B+ down the drain on metaverse), though they have made some good acquisitions.