An inside look at how Google's Ethical AI team fell apart following the contentious departure of Timnit Gebru and then Margaret Mitchell
She was a star engineer who warned that messy AI can spread racism. Google brought her in. Then it forced her out. Can Big Tech take criticism from within?
WiredTom Simonite
Context & Ripple Effects
The Wired piece closes a six-month arc that began when Timnit Gebru was pushed out of Google's Ethical AI team over an email to colleagues in December 2020, followed by Sundar Pichai's "de-escalation" memo that Gebru herself read as casting her as an "angry Black woman." Reporting later showed the team was in turmoil long before the ousters, with friction over how harassment and bias allegations were handled internally.
Margaret Mitchell's departure removed the second co-lead, and by early June current members told reporters they had been in limbo for months with serious doubts Google could rebuild credibility — just days before Wired's inside account. The timing matters because Google had just used I/O to make AI systems its centerpiece while leaving questions about its treatment of critical researchers unaddressed.
First-order effects
Google has lost both Ethical AI co-leads — Gebru and then Margaret Mitchell — leaving remaining team members without leadership and, per their own account, in limbo for months.
Google's credibility problem is now internal as well as external: current staff publicly doubt the company can rebuild trust with its own ethics researchers, on top of the leaked-email fallout from Gebru's firing.
Second-order effects
At I/O, Google chose showcase over accountability, making AI systems a centerpiece while not addressing its handling of dangerous-AI research — a response pattern competitors can copy rather than correct.
Recruiting for internal ethics roles gets harder across Big Tech: the visible fate of two prominent co-leads signals to prospective hires what happens to internal critics, pushing independent scrutiny toward academia and nonprofits instead.
Third-order effects
If the pattern holds, internal ethics teams at major AI companies become structurally subordinate to product timelines — hired for legitimacy but constrained when their findings threaten launches — shifting real oversight to regulators and external auditors by default.
Gebru's post-Google profile, including her continued public commentary on bias and Big Tech power, suggests dissenting researchers increasingly build influence outside the lab rather than within it.
The trend: Corporate AI ethics units are being tested against commercial priorities, with the departure of internal critics like Gebru marking the point where in-house oversight loses independence.
“Gebru had disagreed, claiming she could nudge things, just a little, toward a more beneficial path. “I kept on arguing with her,” Gebru says. Now, she says, she concedes the point.”😐
This (long) article is a great explainer on how the nascent field of AI ethics got overrun by corporate money and power—and how, despite all the PR about responsible and ethical tech, and diversity and inclusion, it's just been business as usual at Google https://www.wired.com/..…
“I'm not worried about machines taking over the world, I'm worried about groupthink, insularity and arrogance” in the people training them https://www.wired.com/...
“She's been thinking back to conversations she'd had with a friend who warned her not to join Google, saying it was harmful to women and impossible to change. Gebru had disagreed...Now, she says, she concedes the point.” A must-read feature by @tsimonite https://www.wired.com/...
“We need more support for external work so that the choice is not ‘Do I get paid by the DOD or by Google?’” Excellent piece on @timnitGebru's exit from Google by @tsimonite: https://www.wired.com/...
5/ The Ethical AI team's primary job was not to conduct research; that was one of our jobs. Our former employer appreciates both research and product impact from research teams. And so, for product impact, I particularly focused on operationalization, as did some of the team.
6/ ""Timnit's behavior was very far outside the norm," Thats what happens when the company hires ONE Black woman to be a research scientist ever and can't even deal with that.
1/ I didn't “endure” rounds of meetings with lawyers abt my first paper. I eagerly sought them out. My role was always hand-in-hand with my former employer, including deeply appreciating, & working on, the tensions lawyers highlighted. I wanted to learn from them. (And still do.)
4/ Its interesting to see Google's responses and STILL learn something new. NOW they're saying: “One reason managers were not more open in explaining their feedback to Gebru, according to Google, was that they feared she would spread it around inside the company.” Yea @JeffDean?
I continue to appreciate journalists [trying to] tell the real story of how and why Google did this to us. @tsimonite has written a more in-depth one. I won't speak for @timnitGebru , but I should clarify a few things before this narrative permeates... https://www.wired.com/...
“There was this hope that some level of self-regulation could have happened at these tech companies,' Raji says. 'Everyone's now aware that the true accountability needs to come from the outside...'” https://www.wired.com/...
1/ Thank you @tsimonite for working on this story since December. I'm not sure how many people have read my PhD thesis but you're definitely one of the few. I know this story was a huge undertaking & done with a lot of research so I won't nitpick much. https://www.wired.com/...