Fully responsible, trustworthy technology is an almost impossible mandate in a tech landscape that prioritizes speed — but that doesn’t mean some companies aren’t trying.
On the heels of the Trump administration’s national AI legislative framework on March 20, in which “winning the AI race” remains paramount, tech developers face tension between the common ethos of moving fast and breaking things versus strategically implementing responsible tech frameworks from the start.
Getting ahead has, in many instances, taken the driver’s seat, the cost of which has become clear. Microsoft’s self-admitted realization that AI-generated code often forgoes accessibility makes human oversight and iteration a must.
For Jenny Lay-Flurrie, who became head of Microsoft’s Trusted Technology Group in February and has worked in accessibility for much of her 21 years with the company, the responsible development and deployment of tech is two-fold: “How do we make sure that we build it right? And how can we make sure that it stays right?”
Microsoft launched its Trusted Technology Group in early 2025 and has since consolidated all responsible tech initiatives under its umbrella, including Lay-Flurrie’s former directive of accessibility.
While Microsoft has centralized its responsible tech under a top-down model, competitors like Google maintain a more engineering-led architecture guided by its core AI principles and specialized safety councils. Techniques vary across big tech, but Microsoft’s approach is one that’s been reshaped since 2002, when Bill Gates released the Trustworthy Computing memo that prioritized things like reliability over new feature development.
AI’s woes (and the people who fix them)
Lay-Flurrie’s foray into the broader responsible tech space may be recent, but she says that it follows the same general principles she’s used to, including fairness, transparency, inclusiveness and accountability. Microsoft operates on the principle that “people should be accountable for AI” regardless of its outcomes.
That’s why, when Microsoft realized its AI wasn’t accurately representing blind people, her team moved to fix the problem.
“Some of the generated imagery of blind people came back with people wearing these horrible full-on blindfolds,” she said. “These models were being trained on a lot of the material that exists in society. Unfortunately, society is not always the most inclusive place, so there are instances where we have to insert data to train it.”
To do this, Microsoft purchased more than 20 million minutes of multimodal data from Be My Eyes, a nonprofit accessibility platform that blind and low-vision individuals can use for free to connect with live volunteers and AI, giving them audio insight into what they’re seeing. “They had a lot of video material that was taken by blind people of their use with canes and dogs and finding keys in the house, and we anonymized the data by blurring faces and all that so that we could train our models more appropriately on blindness,” said Lay-Flurrie.
This process is robust, but Annie Brown, CEO and founder of Reliabl, a machine learning training software working to minimize bias and maximize performance in AI models, said there’s room for improvement.
“More diverse data is just part of it,” said Brown. “If you don’t pay attention to what’s happening at the metadata layer, which is how those images that were uploaded to your data set are labeled, that itself is going to create bias.”
Despite the AI race that’s changing the world, Microsoft is part of a broader movement of companies publicly sharing their responsible tech learnings. Microsoft Learn is freely available to students, academics and developers and includes training modules on responsible AI principles and more. Brown recommends Microsoft also learn from the smaller social good organizations to see “how they’re bringing inclusivity into AI.”
As for improvement, Lay-Flurrie says it comes with the territory. “It’s listening clearly to the feedback, receiving that, iterating, testing and resolving those within as short of a period of time as we can,” she said.
Humans vs. AI
Microsoft is a top provider of enterprise tech, meaning its very own AI is fueling other companies that often make the decision to cut employees in lieu of advanced solutions. Microsoft itself is part of a broader wave of big tech layoffs, though it’s clarified it’s more of a reshuffling of priorities rather than sheer replacement. The company cut roughly 15,000 jobs in sales, gaming and customer-facing divisions in 2025 and hired new personnel elsewhere with a focus on AI infrastructure.
Even as layoffs continue across industries, Lay-Flurrie says AI is already leveling the playing field for previously marginalized workers, including those who are neurodiverse and disabled.
“The first community to get access to Copilot at Microsoft was our disability employee group,” she said. “For the Deaf community, captioning, transcripts, meeting notes, sign language recognition, that gives independence. You don’t have to wait for a cartographer to be there to transcribe what’s being said.”
For the neurodiverse community who received Copilot early on, it helped so much with the cognitive load that “they wouldn’t let me take the license back,” she said.
Diego Mariscal, CEO and founder of global startup accelerator 2Gether-International (2GI), which is run by and for entrepreneurs with disabilities, recognizes Microsoft has made a point to include people with disabilities. “The fact that Jenny’s position even exists at this level is a testament to that,” he said. Even so, including disabled people at the decision-making table is important both from the top down and bottom up. “How can we ensure that, as AI evolves, disabled people are included at the table, not from a charity perspective, but because doing so will ensure that technology and innovation is even more cutting edge and accessible for everybody?”
