Copilot Workspaces is a collaborative environment currently available through Microsoft’s Frontier early access program, but the ability to add personal Autopilot agents to Workspaces will be available soon. At the beginning of the keynote, Nadella said that “Until now, AI interactions have been primarily single-player. Workspaces makes this a truly multiplayer experience, where people and agents move toward a common goal, with a shared context.”
Governance over trust
Trust, the third pillar of Microsoft’s new frontier ecosystem, was something Nadella continued to return to during his presentation, and not in the way some might expect. “We are in a place where alignment is elusive. Mechanistic understanding of these models is elusive. The question is: How can you even trust all this powerful technology? The reality is, by definition, you can’t,” he said.
Breaking away from the norms of corporate advertising sentiment that “AI is magic,” Nadella admitted that the inner workings of AI thought processes remain difficult to understand and that models cannot be trusted to govern themselves, regardless of whether their responses are accurate. Nadella described the solution as “an architectural imperative to separate the provision of intelligence” from the authority given to it. Microsoft says each Autopilot agent will have a unique ID, and administrators can restrict permissions or block it entirely. All agent activities are logged for human review, and Autopilot may request human approval before performing certain actions.
It is said that human ambition, intelligence and trust are the three pillars of the border ecosystem.
It is said that human ambition, intelligence and trust are the three pillars of the border ecosystem.
The promised agent checks and admission that AI systems can’t always be trusted are significant, but Microsoft is asking companies and users to trust the company’s own safeguards, although the effectiveness of those safeguards against errors, dishonest actions or data exposure will still have to be demonstrated. In Microsoft’s view, the solution to untrusted intelligence is not to make AI agents trustworthy, but to control what they are allowed to do. Companies will have to decide how much autonomy they delegate to agents.
Along with the guardrails, Microsoft announced what it calls the “AI Transparency Initiative.” The effort has partnered with more than 40 major companies, including Adobe and Mastercard, to develop industry-wide open standards for AI governance and security. Nadella described the initiative as a way for the industry to “help establish an open standard that makes AI safer, more transparent and more accessible.”