He believes it could have saved lives -- but isn't that true of all weapons?
When thousands of Google employees protested Project Maven -- a Pentagon defense contract for image recognition tech that could have been used for drone strikes -- the company claimed its work was "specifically scoped to be for non-offensive purposes."
But the ex-Pentagon official behind Project Maven has just admitted it might have harmed people after all.
"I fully agree that it might wind up with us taking a shot," former Deputy Defense Secretary Robert O. Work told Bloomberg, arguing that the potential for saving lives might outweigh the concequences.
"They say, look, this data could potentially, down the line, at some point, cause harm to human life," he added. "I said, yes but it might save 500 Americans or 500 allies or 500 innocent civilians."
Work told the publication he was "alarmed" that Google has decided to stop working on Project Maven for that reason, though last we'd heard Google hadn't actually canceled its contract.
Though Google CEO Sundar Pichai recently promised employees that "we are not developing AI for use in weapons" -- it's part of a giant new ethics memo you can read here -- the company reportedly won't ditch Project Maven until the contract expires in March 2019.
A recent report suggested that Project Maven isn't the only government contract that caused concern inside Google.
Google and the Department of Defense didn't immediately respond to requests for comment.