Image source, Alvaro Alvarez/ BBC Image caption, Meredith Whittaker believes science doesn't support the idea that AI will turn against humans
Warning: This article contains strong language.
Meredith Whittaker has little patience for the AI doom that dominates headlines - she thinks warnings of AI turning on humanity is a fear "not anchored in science".
"The risk is very real," says the boss of the Signal Foundation, which runs the encrypted messaging app of the same name, "but it's not the shiny ghost stories the wild-eyed boys are talking about".
Instead, she believes the more immediate danger is the technology being used carelessly.
She cites an alleged incident, reported by CNN, in which a chatbot used by US intelligence invented information about a Chinese vessel carrying nuclear material, reportedly leading the US to prepare a military response.
"We almost went to war with China," she tells BBC Global Women.
"A system that produces faulty intelligence and almost gets us into a war with China should not be considered super-intelligent by any standards."
Signal has an estimated 40-70 million users - a drop in the ocean next to WhatsApp, which has billions.
But it is renowned for its tough stance on protecting users' privacy - it hides who is sending a message from its own servers, limiting the data available to outside observers or hackers.
Time magazine named Whittaker one of the most powerful people in AI. She moves through the rooms where its future is discussed, from Silicon Valley boardrooms to Downing Street, Davos and the Vatican.
She studied literature at university and joined Google in 2006 as an entry-level temp with no computer science degree.
In late 2017, she made her name when a colleague told her about Project Maven, a secret Google contract with the US Department of Defense to build AI for drone targeting.
She understood what it meant to "compress the kill chain": to shrink the time between a machine flagging a target and someone deciding to fire, reducing deliberation. With thousands of colleagues, she pushed back.
"This is going to launder bad intelligence and bad data," she remembers arguing publicly.
The following year Whittaker helped organise a walkout, in which some 20,000 staff protested against the company's handling of several issues, including what they felt was the rushed introduction of new technology. Google dropped the contract and Palantir picked it up.
Image source, Alvaro Alvarez/ BBC Image caption, Whittaker criticises venture capitalists' lack of support for startups led by women
She draws a direct line from Maven to the US missile attack on a school in Iran this year, which she says relied on the system and on outdated intelligence that identified the school as a military target.
"This is exactly what we were warning about in 2017," she says.
In response, Palantir told Global Women there was "no evidence that Palantir's software was at fault in this terrible tragedy".
It added that "Palantir is not responsible for the underlying data nor identifying intelligence deficiencies". And earlier this year it's UK and Europe boss, Louis Mosley told the BBC that "there's always a human in the loop, so there is always a human that makes the ultimate decision".
The US military has yet to release the findings of its investigation into the Minab strike, but Defense Secretary Pete Hegseth has said it would "never target civilian targets".
Government…