- Ex-xAI and Google engineer Tina Oberoi is in early talks to raise $50m for Moir, a Palo Alto startup that helps people prove they’re not a deepfake
- The round would value the unlaunched company at $250 million.
- She’s the latest of several former xAI staffers spinning problems the frontier labs helped create into standalone startups
Tina Oberoi spent eight months at xAI working on the data behind Grok 3 and Grok 4. Now she’s building something aimed squarely at what that kind of model makes possible: content so good that nobody can tell whether a real person is behind it.
Oberoi is in early discussions to raise roughly $50 million in seed funding for Moir, a Palo Alto startup that aims to give individuals a way to prove their identity online, according to a Bloomberg report citing a person familiar with the matter. The talks would value the company at $250m.
That is, notably, everything anyone knows about Moir. There’s no product, no customer, no public website, and no funding announcement — which makes this a story mostly about who’s backing it and why that’s enough to draw nine-figure interest before a single feature has shipped.
A different kind of identity problem
Most identity verification is built the other way round: a bank or a marketplace buys enterprise software to check that you are who you say you are. Moir is reportedly flipping that, aiming to let individuals hold and prove their own identity credentials rather than relying on whichever company happens to be checking them. Whether that means deepfake detection, identity credentials, or straight authentication hasn’t been said.
Before xAI, Oberoi spent more than five years at Google, moving from software engineer to customer engineer on data and AI. She’s one of at least eleven people who’ve left xAI’s founding and early ranks this year to start their own companies, a pattern that’s become its own mini-genre of 2026 funding news.
Crowded corner of a hot market
The identity-and-trust space isn’t short of capital right now.
Tech Funding News has covered Paris-based Prelude’s $20 million Series A, which built a continuous trust score out of telecom and behavioural data rather than a one-time check. Polish e-ID platform Authologic raised $8.2 million to attack the same document-fraud problem from the compliance side. Seattle’s Vouched raised $17 million to extend identity checks to AI agents, not just humans. And Deel folded deepfake-detection into its hiring platform by acquiring Clarity for an estimated $40m to $50m in August — the same order of magnitude Moir is now trying to raise for an idea with no product yet.
Where Moir would differ, on the thin detail available, is the individual-first angle: identity as something a person owns and carries, rather than something a company verifies on their behalf.
The xAI alumni startup wave
The bigger draw for investors may be less the product than the pedigree. River AI, founded by xAI co-founder Igor Babuschkin, raised $1.1 billion within months of incorporating. Humans&, founded by researchers out of xAI, OpenAI, Anthropic, and Google, raised $480 million at a $4.48bn valuation. Moir’s reported raise is a fraction of either, but it’s the same trade: frontier-lab alumni turning the problems their old employers created into venture-backed companies.
Whether that pattern holds once these companies actually have to ship something, rather than trade on where their founders used to work, is the question the next few months of this space will answer.
