Google has fired both leads of its Ethical AI unit, leaving doubt about the direction of the team. The company has come under fire in recent months for alleged gender bias, racism, gaslighting and censorship.

Last month, the company's workers formed the first major union in Silicon Valley in support of employee activism. The most recent controversy involves the firing of Margaret Mitchell, co-lead of Google's Ethical AI unit and an ethical AI expert. Mitchell specialized in a variety of topics, including machine learning bias, race, gender diversity and language models for image capture. She announced her firing in a tweet last Friday.

The move follows the firing of Timnit Gebru in December. Gebru was the other co-lead of the Ethical AI team, and had worked with Mitchell in that capacity for about two years. Google said in a statement that Mitchell had been fired after confirmation of "multiple violations of our code of conduct, as well as of our security policies, which included the exfiltration of confidential business-sensitive documents and private data of other employees."

Gebru claims she was fired for criticizing the company for silencing marginalized voices. She also co-authored a research paper urging tech companies to ensure AI language systems do not promote gender bias. Google says Gebru resigned.

Sundar Pichai, CEO of Google, has publicly apologized for the handling of Gebru's removal and has promised to work on restoring employee trust. Gebru's removal has been cited as one of the reasons workers created the groundbreaking Alphabet Workers Union last month.

Alex Hanna, a Google employee who worked closely with Gebru and Mitchell, tweeted that the company has been running a "smear campaign" against the two researchers. Google has not commented on her remarks.

Google said the company had conducted an internal investigation into Gebru's departure that will lead to a change in policies. Google said changes will include new procedures surrounding "potentially sensitive employee exits" as well as linking pay to diversity goals in some cases.