Microsoft cuts its AI ethics team as the race heats up

Picture racing down an uncharted mountain road at full speed, and the driver decides the best way to move faster is to tear out the brakes.

That is essentially what just happened inside Microsoft. On March 14, reports confirmed that the company eliminated its entire ethics and society team within the AI organization as part of its ongoing 10,000-person layoff wave.

This was not a team of paper pushers. At its peak in 2020, the group had around 30 engineers, designers, and philosophers. They built practical toolkits like Judgment Call, an internal role-playing game designed to help developers anticipate how their code might actually harm people before shipping it.

Sausage-making at breakneck speed

Eventually, that friction became intolerable for company leadership determined to win the AI race.

Back in October, Microsoft slashed the team down to about seven people. According to leaked audio from a meeting at the time, corporate vice president John Montgomery made no secret of where the pressure was coming from. He told the team that CEO Satya Nadella and CTO Kevin Scott were applying very high pressure to get the latest OpenAI models into customer hands at breakneck speed.

When employees pushed back, Montgomery shut it down. "There’s a lot of stuff being ground up into the sausage," he told them, adding they should probably be thankful they didn't have his vantage point.

On March 6, the remaining members were pulled into a sudden Zoom call and told their jobs were gone. Microsoft claims it still maintains its Office of Responsible AI to craft corporate principles, but former staff say the real loss is the bridge between policy and actual product engineering. Abstract rules mean nothing if there is no one left to show developers how to apply them.

A paper trail of ignored warnings

The team's final battles centered on Microsoft's rushed integrations with OpenAI.

Last year, the ethics group sounded the alarm on Bing Image Creator, powered by OpenAI's DALL-E. They found that typing a living artist's name produced outputs practically indistinguishable from the original artwork. Ethicists urged Microsoft to block living artists' names as prompts and build marketplaces to compensate creators. Those mitigation strategies were ignored, and the tool shipped to pilot markets anyway.

To executives looking at spreadsheets, caution looks like a liability. Microsoft has poured $11 billion into OpenAI, and Nadella knows that peeling just 1 percent of search market share away from Google translates to $2 billion in annual revenue. With Bing recently crossing 100 million daily active users, leadership felt justified in running full steam ahead.

Meanwhile, copyright lawsuits from the likes of Getty Images are already piling up across the industry, proving the team's warnings were right on the money.

Why it matters

We think Microsoft is taking a massive, short-sighted gamble here.

Keeping a high-level policy office while firing the people who sit alongside engineers to catch real-world harms looks a lot like performative compliance. For everyday users and creative professionals, it means tech giants are willing to treat copyright infringement and hallucinated content as external collateral damage rather than product flaws.

Across the industry, this sets a toxic benchmark. When Google ousted ethics researcher Timnit Gebru in 2020, it sparked massive internal outrage and tarnished its reputation. Microsoft seems willing to absorb that exact same hit because it believes beating Google to the search prize is worth whatever PR fallout follows.

Moving fast and breaking things is an old playbook, but applying it to generative AI with no internal safety check is playing with live ammo. Ripping out the brakes might shave seconds off your lap time, but it almost guarantees a catastrophic wreck down the road.