Chairwoman Maloney, Ranking Member Jordan, and members of the Committee, thank you for inviting me to speak today. My name is Meredith Whitaker and I'm the co founder of the AI Now Institute.
In twenty twenty, Meredith Whittikue gave testimony to US Congress about the threat of artificial intelligence.
Facial recognition poses serious dangers to our rights, liberties, and values. Whether it's used by the state or private actors, the technology does not work as advertised. Research shows what tech companies are.
She turned her back on a career at Google and wanted to send the public message.
Facial recognition relies on the mass collection of our biometric data. It allows government and private actors to persistently track where we go, what we do, and who we associate with.
But she didn't turn her back on tech entirely.
If you care about the over policing of community, communities of color, or gender equity, or the constitutional right to do process and free association, then the secret of unchecked deployment of flawed facial recognition systems is an issue you cannot ignore.
From Schwartz Media, I'm Ashlin Macgee. This is seven AM. Meredith Whittaker is one of the world's most influential people in AI, except she doesn't work in it anymore. She spent years deep in the bowels of Google until a secretive defense contract changed.
Her entire perspective.
She's now the president of Signal, an encrypted messaging app used by journalists, whistleblowers, drug dealers, militants, and really anyone who.
Wants to keep their communications secure.
Privacy advocates love it, law enforcement hate it, and Meredith Whittaker is fighting governments desperate to regulate it. Today, the President of the Signal Foundation, Meredith Whittaker a head of her public appearance at the Wheeler Center in Melbourne on the tech giant to rule the future and why she says signals not one of them.
That's after the break, Meridith.
A lot of people would describe the moment that we're in now when it comes to artificial intelligence is a really pivotal one. But I wanted to ask about you personally and what led you to begin researching and examining AI.
Yeah, well, it boils down to my having been in the tech industry now for almost twenty years and a little over ten years ago or so beginning to see AI or machine learning as we called it, then emerge as a heart new thing inside Google where I worked, and starting to ask questions about what AI was and.
How it worked.
That led me to a more critical perspective on these set of technologies and actually to understanding their deep inextricable connection to the platform surveillance business model.
So talk to me a little bit more about that.
What were you working on at Google at the time, and how did you see it come up in your workplace.
Yeah, so I'd done a number of initiatives at Google by that time, and one of the big projects that I kind of cut my teeth on in tech was an effort called Measurement Lab whether we can trust data or not, whether it should inform our decision makings, or
whether we need to be a little bit wary. So I was coming into this AI moments around the early twenty tens with that type of skepticism, and of course when I saw that these technologies we were calling AI, these machine learning models were being trained, were being informed by much less robust data, often data that was making very socially sensitive claims about people and population, data that was created in service of selling ads or you know,
tracking clicks and preferences across browsers or the search instance or what have you. That really animated my concern around AI. There was a moment where I realized that there were some real high stakes to what I was seeing, you know, and potentially those stakes were beyond what I had understood before.
And this was.
Twenty fourteen when a research group from Harvard was presenting to myself and some colleagues an initiative that aimed to create an AI system to predict genocide. And I remember some of thinking through what they were doing, and you know, asking some pretty basic questions, right, Like, the term genocide is not clearly defined. It is a politically contested term. So I was asking questions around, Okay, well, how do
you define genoide? What type of data do you rely on to model predictors for genocide?
Right?
So you know, where are you getting this? How do you take responsibility for the fact that if you deployed such a predictor you may be putting your thumbs on the scale, You may be in fact inflaming or otherwise shaping volatile political context outside of your scope of awareness.
And I got no good answers to those questions.
So you stayed on a Google after that? Then what happens?
Yeah, twenty seventeen.
Someone sent me a signal message one night in the fall of twenty seventeen saying, Hey, I think you need to know this. I happen to privilege access to the small team of people who are working on this secretive contract with the Defense Department to build machine vision and surveillance systems for the US drone war, you know, which had been deemed illegal by every blue chip human rights organization.
So I was extremely angry, and I was very worried because this was Google, who is a company that had effectively been enabled by regulators and governments around the world to take on incredibly significant roles in our society.
Right, you know, you no longer go to the library, You google it.
Right.
They are positioning themselves as the gateway to the world's knowledge and understanding.
They are, you know, creating.
The platforms for human interpersonal communication. And so we had given this power to these companies on the basis of very clear promises to do the right thing, to not be evil, to not misuse this power. And this was crossing a red line in my view.
Artificial intelligence drones warfare and Google. It's a mixture that caused an uproar inside the tech giint, where the early motto was don't be evil. So what's behind Google's contract with the Department of Defense for a project called mabn me Now from.
Google, who was now yoking their fortunes to the world's most lethel military in the context of a drone war that was again deemed illegal, that was massacring wedding parties based on inaccurate data streams, and that was setting a precedent for a kind of technologically enabled empire that I found really dangerous and morally repugnant.
Is that why you left Google?
No, That's why I started organizing at Google. That's why I started rallying the people around me to push back. That's why your thousands of people along with me, wrote letters, gave out literature, signed petitions, went to the press, spoke publicly against this contract, and ultimately forced Google to cancel it.
A letter to Google CEO Sunder Pitchai signed by more than three thousand Google workers.
Here's what it says.
Quote, we believe Google should not be in the business of war. Therefore, we ask that project may then be canceled and that Google draft public publicized, and.
So you know, no, it's not where I left.
It's why I started fighting in a much more concerned way and started relying on.
Techniques from the past of you know.
Kind of building power among workers and stakeholders to push back on corporate decisions. Because it became very clear that you change from the insight doesn't happen simply by making a good point.
After the break the harmful content on Signal that Meredith whittyk says is being weaponized. So, Meredith, you now the president of the Signal Foundation. What was it about Signal that appeals?
Why join?
Oh?
I love Signal. I've been a Signal super fan since day one. I've been friends and admirers of the Signal folks. I think there is a future in which Signal sustains and grows and continue the incredible momentum that it's already generating.
And that is a future where we can speak privately, where journalism is possible, where human rights work is possible, where dissent is possible, where we can where activists and dissidents and militaries and CEOs and anyone with something important to say that could be weaponized against them by their adversaries or those in power. Can use signal to truly communicate privately, and that's the future that I think we need, particularly given so many gnarly issues in the world, particularly
given the increasing prominence of authoritarians and government. We must preserve the ability to communicate in a truly private way because we cannot have all of our thoughts and our correspondence subjects to mass surveillance and then weaponized by those in power.
Meredith, you've been in Australia now for a little over a week. There's a big push here for our government to improve regulation of the Internet. What do you make of it?
You know, I'm learning and listening to a lot of folks in Australia. I think you know, I was really heartened to see that the E Safety Commissioner released language that you know, looks like a great first step in protecting encryption. I think Australia has a big role to play in these debates.
We will be seeking cooperation wherever we can, but we need to take action and the online.
There is one really gnlly issue though that signal is grappling at the moment, and that is this big conversation around how encryption and platforms that are encrypted, including signal, are being used to share terror and child exploitation content.
That must weigh on your mind.
Well, weays on my mind that people are weaponizing such socially significant and often monstrous issues to attack the one technology we have that actually enables private communication. Like Look,
this is a long standing tension. The attack on encryption, the attack on the ability to communicate and transact privately over networked computation, has existed since the mid seventies, when public key encryption technologies were invented and security services in the US literally tried to stop publication of the paper that documented these systems. Then you go to the nineties
and these same tropes emerge over and over again. You have a nineteen ninety two op ed and Wired by then NSA head Stuart Baker that claims that these very rudimentary encrypted systems are the providence of terrorists and pedophiles and that no one beyond those demographics should really be concerned about privacy because the rest of us don't have anything to hide. Now, the landscape of surveillance in the seventies, in the nineties, and now is vastly different, but somehow
the arguments never change. Somehow it is always the state must have access to every realm of human life in service of rooting out the bad guys and supporting the good guys. But of course that is not how it works in practice, and what we see today, in my view, is a renewed attack on the ability to truly communicate privately, based on emotional claims that are not supported by data. There is no evidence that encryption is a causal factor in child abuse or terrorism.
I'll just jump in their meritis because I just want to explore that a little bit more. And I guess what we're talking about is not whether it's a causal factor, but the fact that this type of content is shared on encrypted platforms. It does get shared on encrypted platforms. I think that that's fairly obvious these it do you accept, then.
Well, the purpose of encryption.
Is to provide provable privacy. We cannot look into the messages that you send on signal, so I'm sure a number of things are shared on encrypted platforms, but there is no evidence that encrypted platforms produce more abuse. You know, there are things that we do to ensure that the harm surface on signal is reduced. The primary one is
we are not a social networking platform. But again that framing then immediately implies that somehow, if there were, if platforms, if tech companies just did the right thing, the deep sickness of child sexual abuse would somehow go away. And I think, to me, that's really dangerous because we've abstracted the issue of child sexual abuse into an online problem that a couple of engineers turning some knobs and giving law enforcement and potentially authoritarian governments in the US extra
access will somehow solve. I don't think one person who showed up in Epstein's Black Book has been prosecuted, you know, I do need to say is incredibly dangerous in the US right now, there's a woman living in prison this moment because Facebook turned over messages between her and her daughter that documented their excess of reproductive care and dealing with the aftermath. In the state of Nebraska after the Dobbs decision, we are learning more.
Now about how Facebook turned over a mom and her teen's daughter. Her teenage daughters chats to police in Nebraska court document show police served a Facebook a warrant as part of an investigation into an alleged illegal abortion. The probe started back in a year.
We need to demand the kind of precision and rigor that is often difficult in the face of highly emotionally charged topics.
But there's no way to get this right if we don't do that.
Finally, Meredith, there are AI evangelists and AI doomsdays in this world.
What do you identify as?
I am an AI realist, which means I'm not clear that AI is what we should be interested in so much as the configurations of power that are going to use AI and the way that AI could be used to generally harm the most marginalized.
I'm not really a user of AI, putting aside that I.
Don't really use GPT or these sort of generative toys. I think in most cases, like most other people, I am more the subject of AI than a user of AI. And what I mean by that is, you know, well, all walk down the street and a CCTV camera will pick up our face, and that CCTV camera may transmit that biometric data to a clear view AI facial recognition database that then pulls up a profile of us.
Right I walk into a bank, I might apply for a loan.
An AI system that is a licensed from Microsoft on the back end may be involved in making a determination about whether I get that loan or not. But in none of those cases am I actually using AI. I am conscripted as an AI subject by the institutions and governments that have chosen to implement AI, and so I think we need to recognize that power dynamic when we think about using AI.
This is generally a tool of.
Those who have power that's used to subject those who have less.
Thanks so much for your time, Meredith.
Thank you, it's great to be here.
Also in the news today, Julian Assange's walked free from Belmarsh Prison and will face court today on the island of Saipan, a US commonwealth in the Pacific Ocean. Under a deal struck with the US Justice Department. Assange is expected to be sentenced by a judge to time already served,
and then he'll be free to return to Australia. Prime Minister Anthony Albanezi revealed Assange would be a company to court by the current UK High Commissioner, but said he wouldn't comment further until after a final judgment's being secured.
We want him brought home to Australia and we have engaged and advocated Australia's interests using all appropriate channels to support a positive outcome, and I've done that.
Since very early on in my prime ministership. I will have more to say when these legal proceedings have concluded, which I hope will be very soon, and I will.
That's all from the team at seven am for today. My name is Ashlin McGhee. Thanks so much for your company. See you again tomorrow.