As a cofounder of Clearview.AI — this’ll be litigated soon enough — I pay careful attention to a lot of the developments in the facial recognition debate and generally to how technology is sold into the government.
There are several bad faith arguments made against facial recognition which we will explore but first some back story. I’ve talked about Silicon Valley’s traitor problem before but it really does deserve some repeating. Silicon Valley has a problem when it comes to foreign sources of capital and that, in turn, affects the sorts of tech deals which can and can’t be done.
The money, in other words, doesn’t want to help with our national defense. The Americans holding that money oftentimes are the witting agents of the foreigners who gave it to them.
Naturally I kept careful receipts about who did and who didn’t invest in Clearview.AI and what reasons they gave and what I suspect the real reasons are. Though I might be seen as vindictive I think it’s important to understand who believes what. It’s important for everyone to show their cards.
This is particularly true as there’s been considerable money expended in targeting and misinforming the public about how Silicon Valley works. Young people are often misled by Silicon Valley’s propaganda about what their true intentions really are. I have elsewhere called these intellectual cul de sacs. For what it’s worth I strongly opposed going to Silicon Valley for capital but one of my other cofounders — Hoan Ton-That — overruled me. I wanted to go to Northern Virginia cash for — ahem — obvious reasons.
Why would Sequoia, backed as it is by China, invest in Clearview.AI, its American competitor? Doug Leone, Sequoia’s partner, loved the product — he couldn't tell some of his Asian employees apart apparently (yeah, seriously) — and still refused to invest. Sequoia scout Jason Calacanis even interviewed Hoan on his fake podcast but ultimately didn’t invest. Could it have been his family’s longstanding ties to the mob? I wonder.
Sequoia is now rightly being looked at by U.S. intelligence for their very close ties to China. Good. (And about time!)
Did you know that Sequoia backed SenseTime which is banned by US government? Sequoia partner Michael Moritz recently complained about San Francisco but San Francisco was one of the first places to ban law enforcement technology, including facial recognition.
Founders Fund partner Trae Stephens opposes facial recognition on “ethical grounds” — even though he had invested right alongside Jeffrey Epstein in Israeli spyware technology company Carbyne.
Interestingly Stephens’s arguments against facial recognition hewed closely to the arguments of Israeli firm Oosto (formerly Anyvision) which realized it was way behind Clearview. Stephens’s mentor, George Tenet, was drummed out of CIA for his close ties to the Israelis and for his comments on WMDs were a “slam dunk.” Rather interestingly Stephens spent time in the Gaza Strip as a boy doing “mission work.”
I actually agree with Stephens about the importance of meeting the threats that other countries pose and of having a seat at the table.
He writes on the ethics of investing in defense tech:
In debating the value of our investment in defense technologies, we cannot ignore the fact that if we allow others to build these technologies while we stand idle, we will lose the power to regulate their use, we will allow aggressive autocratic regimes to take the lead, we will voluntarily limit our power to deter harmful conduct (including genocide, repression, and interference with international norms), and we will cede to the most belligerent and authoritarian states the power to impose insidious legal and moral standards on the US and its allies without consequence.
This is happening presently inside of the United Nations’ International Telecommunications Union (ITU), where China is hard at work leading the development of regulatory standards for the use of facial recognition technologies, which they have already exported to dozens of countries around the world. This is not a hypothetical future threat — it is a real and present one.
So why is Stephens AWOL on this topic?
Cartelizing the investment flows into the defense sector is what Stephens does but he seemingly totally missed the growth of satellite tech companies.
In any event Stephens recently told Kara Swisher that “I happen not to believe that you cannot[sic] deploy facial recognition ethically so we won't do it. It doesn't really make sense.”
I happen to believe that a lot of people are worried about the kind of detections that facial recognition would present and that when properly deployed it’ll render in person informants and spies a thing of the past.
Even my friend Palmer Luckey gets this wrong when he told the Financial Times that “Anduril did not draw the line at building weapons but would not use facial recognition software, because the technology could never be reliable enough to trust in life-or-death situations.”
Where Palmer has failed to go other defense tech has gone. Vice has the story: “US Military Signs Contract to Put Facial Recognition on Drones.”
The Commerce Department recently released its list of NIST vetted algorithms in facial recognition. Unsurprisingly Clearview took top marks once again. You can see for yourself.
When you’re at 99.48%+ of accuracy we are talking levels of accuracy that few tech companies could match. Could OpenAI or Tesla meet that level of scrutiny? I think not. And besides, facial recognition is a law enforcement tool and part of an investigative process.
You should always keep the human in the loop for decision making — and the government for assessing claims by tech companies.
We should use NIST to set rules for all technology sold into the federal government. Indeed having NIST set the rules could be a standard for all sorts of defense technology. I’ll always take the Pepsi challenge of whether or not companies I invest in or start are the best in class.
But having the best technology doesn’t mean you have the best leadership and indeed sometimes the best technology won’t be funded simply because of the jealousy.
Naturally I rescued the company by securing investors through my network—only to have Hoan try to push me out of the company for talking to the press after he was lying about me to the press. I’m suing and it’s going to be a whole thing even though it didn’t need to be. Hoan just wanted to protect his ego and not the shareholders or the company. Many such cases!
But truthfully I don’t support facial recognition only going to law enforcement. I think that in a world where black men are routinely killed by police that can only end in brutal policing.
It’s my contention that just as the Chinese cash needed to be swept out of satellite space so too does Chinese cash need to be removed from the facial recognition sector.
Let me suggest that there are a few options here that we might explore:
Facial recognition should be auditable after a period of time.
Facial recognition should be licensed, especially for the responsible public and subject to random audit. Perhaps we could even have a jury pool selected from the public?
To the extent that AI cannibalizes publicly available stuff, it should only be done for public safety considerations. It’s wrong to steal the style or work of artists.
We have an obligation to develop the best technologies with the best people running them.
You're just a background character, Charles, and don't take that the way you might be thinking. Clearview AI, as well, was just one more 1 or 0 to something else.