Google has made responsible AI development a stated priority for 2024, but an internal change has raised questions about how the company will police its own AI work. The company’s Responsible Innovation team, known as RESIN, has lost its founder, been divided, and had most of its staff moved into trust and safety, according to four people familiar with the changes.
The company says the shift will help scale responsible innovation across Google. Some employees, however, are waiting to learn what the new structure means for AI principles reviews, especially as Google pushes harder into generative AI.
What changed inside Google
RESIN sat inside the Office of Compliance and Integrity, within Google’s global affairs division. Its role was to examine internal projects against Google’s AI principles, which guide how the company develops and uses artificial intelligence.
The team was small but important. It conducted over 500 reviews last year, including work related to the Bard chatbot, according to an annual report on AI principles work Google published this month.
RESIN’s founder and leader, Jen Gennai, left her role this month. Her LinkedIn profile lists her as an AI ethics and compliance adviser at Google as of this month, which sources cited in the source article said may suggest she will soon leave the company based on prior departures.
Another influential member, Sara Tangdall, also left this month. Tangdall had been lead AI principles ethics specialist and is now responsible AI product director at Salesforce, according to her LinkedIn profile. Tangdall declined to comment, and Gennai did not respond to calls for comment.
How RESIN is being split
Google divided Gennai’s team of about 30 people into two groups, according to the people familiar with the change. Company spokesperson Brian Gabriel said 10 percent of RESIN staffers will stay in place, while 90% were moved to trust and safety.
Trust and safety also sits in the global affairs division and works on abuse of Google services. No one appears to have been laid off, according to the sources.
What remains unclear is the practical design of the new review process. The rationale for the split and the division of responsibilities could not be learned. Some sources said they had not been told how AI principles reviews will work in the future.
Gabriel declined to explain how RESIN’s AI project review work will be handled going forward. He framed the restructuring as evidence that Google is investing in responsible AI development, saying the move “brought this particular Responsible AI team to the center of our well-established trust and safety efforts, which are baked into our product reviews and plans.” He added: “It will help us strengthen and scale our responsible innovation work across the company.”
Why RESIN mattered
Google created the Responsible Innovation team in 2018 after AI experts and others at the company publicly protested Project Maven, a Pentagon contract that used Google algorithms to analyze drone surveillance imagery. After those protests, Google introduced AI principles saying it would use AI to benefit people and never for weapons or for undermining human rights.
Gennai helped author those principles. RESIN became a central steward of them, giving teams across Google a place to submit projects for review.
The team’s influence was not only advisory. It provided feedback and sometimes blocked ideas it viewed as conflicting with the AI principles. The group stopped the release of AI image generators and voice synthesis algorithms that could be used to create deepfakes.
Still, seeking AI principles guidance was not mandatory for most teams. That made RESIN different from privacy reviews, which every project must undergo. Gennai argued that early responsible AI review could prevent costly ethical problems. “If implemented properly, Responsible AI makes products better by uncovering and working to reduce the harm that unfair bias can cause, improving transparency and increasing security,” she said during a Google conference in 2022.
The broader AI pressure
The RESIN changes come as Google is competing more aggressively in generative AI. Sundar Pichai recently emailed employees with company priorities for 2024, placing responsible AI development at the top of the list.
Some employees now worry that faster product development could become risky. Those concerns have grown with the restructuring of RESIN, according to sources cited in the source article.
Google has also been streamlining more broadly. It has spent the past year laying off thousands of workers, shutting down services including its Podcasts app, and cutting features from Google Assistant. The company has focused around a few AI initiatives while trying to deliver advances more quickly to users.
Pichai and other Google leaders have said the company can move faster on AI while staying responsible about possible dangers. Google also joined OpenAI, Microsoft, and several other large AI developers in a voluntary White House pledge to assess societal risks and national security concerns related to advanced AI.
What remains unresolved
RESIN is not the only AI-related group at Google to be reshaped. Last year, Google merged DeepMind with Google Brain, creating Google DeepMind to unify development of foundational models behind tools such as Bard.
For Google’s most advanced AI models, including the recently released Gemini, ethical reviews fall to Google DeepMind’s Responsibility and Safety Council rather than RESIN, according to a technical paper published last month.
RESIN has still influenced Google’s generative AI products. A company report says the team helped trigger a decision to limit Bard from using personal pronouns, in an effort to avoid users treating it like a human.
The central question is now less about whether Google still says responsible AI matters. It does. The unresolved issue is how the company will preserve clear, effective AI ethics review as RESIN’s original structure is broken apart and its future role becomes less defined.