Regarding the CIA Funding 120 AI Startups
Consider that everyone is swooning and fawning and working themselves up into a tizzy about the potential catastrophic outcomes of AI through some Terminator rogue cluster driving it and its fellow datacenters to global dominion over the human bodies on which it feeds — wait a minute, that is literally what the VENTURE CAPITALISTS want to do, not a machine learning algorithm. How we tell on ourselves!!
Anyhow, as we all go batshit about all this FUD (created by OpenAI, btw), AI is being pushed into dozens of weapons, intelligence, surveillance and policing startups. In fact, the CEO of Anduril — the VC’s monster weapons startup — recently made comments indicating that Anduril is in Israel — with their AI lethal drone swarms of course. His comment, at a WSJ business conference: “Israel has my unqualified support. They have the support of Anduril in a moral sense. I obviously can’t talk about what we’re doing over there specifically. That’d be a question for a different day.”
I don’t know, I feel like “what we’re doing over there specifically” implies that there is, in fact something being done over there, specifically, in the first place. If this is indeed the case, this is a direct example of AI being used to kill children in a genocide.
This is actually not new, and you should impress upon yourselves that we, the people, saw the chatbot long after this shit was being used to kill, well, children. From a recent piece I wrote on venture capital collaboration with Israel:
Consider that in May 2021, horrific Israeli aggression against Palestine was called “the first AI war”, with widespread usage of these technologies. This is a full year and a half prior to the launch of ChatGPT, and far after Palantir, with offices in Tel Aviv, was using AI technologies for its sick creepy shit of various kinds. Quote from the IDF on the 2021 assault:
“For the first time, artificial intelligence was a key component and power multiplier in fighting the enemy,” an IDF Intelligence Corps senior officer said. “This is a first-of-its-kind campaign for the IDF. We implemented new methods of operation and used technological developments that were a force multiplier for the entire IDF.”
And then we have the matter of the CIA’s involvement in AI, which, in a lot of the misguided paranoia, is getting left almost completely out. In-Q-Tel is the CIA’s venture capital firm, in Silicon Valley, where In-Q-Tel operates openly, and more openly than ever now that a rogue group of venture capitalists is forcibly steering the entire industry into weapons production; we get to see the machinations of the CIA with a level of insight not afforded in other areas, making it a good dimension for study.
Did you know that the CIA has funded over 120 artificial intelligence startups? They are listed openly on the In-Q-Tel website. And the site also states directly that In-Q-Tel is the CIA’s VC firm — nothing about this a secret and the tech industry has allowed this to happen going back to 1999. These investments mean that the CIA has been one of the most pivotal entities in the development and state of artificial intelligence today, just as they were in web 2.0 and the big data agenda, which necessarily gave rise to the AI age.
Before we walk through some of the AI startups indicated and try to get a sense of what exactly they have funded, let’s lay some context about how we got here. The CIA, like most of the Valley and venture capital it had fused to, spent most of the last bubble collecting data. That was the whole theme of the bubble: Big Data. And the goal was to go and get as much of it as we could, and a huge portion of the tech development at this time was shaped by this goal. During this era, the CIA was spending a lot of time funding, of course, big data startups, and that is how I got to see the CTO of the CIA, speaking at a Big Data Conference in New York, tell a room packed with technology executives and their pions (played in this case by me), that the CIA’s goal was to collect ALL of the data in the world. And store it FOREVER. Making complete sense if you are the worst intelligence agency known to mankind. Now, these revelations are remarkable in particular because they came pre-Snowden, and discussed heavily the use of devices and social media data / consumer surveillance as primary features.
Artificial intelligence is only possible due to the mass accumulation of human data, which venture capitalists and, I guess the CIA, lmfao, have had as the core propulsion of the field since the dawn of social media. Accumulation of data made higher and higher levels of data processing and analytics possible and necessary. Artificial intelligence is fundamentally about reasoning about gigantic data stores, all the bullshit they have harvested from us fills a lot of datacenters and they had a big fucking problem on their hands: how are we gonna use this shit to fuck these people even more. There are a lot of ways to slice and dice huge data stores, there are so many ways that this could be done and developed, and what you see is far from inevitable — there are some very specific choices being made, by venture capitalists, regarding how to do that, what to find out, what to expose, how to do it, how to sell it. It turns out it does actually matter who, what, when, where, why and how, of these technologies. Taking this material tact makes it much easier to get out of nebulous, often confusing and definitely propagandized sphere of the term AI; which is so full of non-meaning that it has become meaningless at this point. So I suggest we frame this in terms of: massive data stores (how they got them, who attained them, how the extraction was done, under what premises, what is done with it), reasoning about those data stores, and how those things are being used by specific people, for specific things.
Now, while 120 sounds like a shocking number and it is — we should be shocked that the CIA has funded this many companies PERIOD and this is only a selection of their portfolio — we have to consider that, like most venture entities in Silicon Valley, the CIA is pumping those numbers a little bit. In fact, a lot of the startups they list are ass-old companies that have long been sold, IPOed, turned into scrap, shut down, sun-setted, turned into a documentation file on the home page of the M&A. This is part of Valley shit where we pump things up to look more impressive than they are, and we tack on some buzzwords of the current thing onto whatever we are holding in our hands at the moment, even if it’s our own dic- . But the truth in this is somewhere in the middle, which indeed, these companies have paved the way into artificial intelligence and many of them have “pivoted” into artificial intelligence, or their data assets and their management tools have found their way into the sphere of artificial intelligence; this speaks to the connection / evolution of big data into artificial intelligence. Some of the bigger names in this category are Gitlab, Databricks, Pubnub, Cloudera and Cloudant; these and other big data building blocks that captured most of their attention in the last bubble.
This points us to the actual origins of artificial intelligence; AI has grown naturally not out of the public’s needs for getting insights from collective data — this is something that has never been available to us. We are shown a tiny surface area of the information, and we are given little consideration in what to do with it. In fact, collective data could be transformative, but we have the opposite of collective data: stolen shit. Absolutely zip. Rather than from us, who gets the dinky chatbot, AI has grown from the needs of venture capitalists, tech giants, corporations, intelligence agencies and the military, to make sense of the huge data harvests they have, the years of collecting and storing as much of our data as possible. We get a shitty chat bot, while they are performing incredible complex technical feats that seek to supercharge the ability of exterior and hostile entities. In many ways, AI was a technical inevitability as the amount of data harvested has overwhelmed their ability to make sense of it all; AI provides this. And further, AI is additionally yet another way of harvesting data, as illuminated by a recent article on OpenAI: “The San Francisco-based startup said Thursday it would like groups and communities to contact it to collaborate on data partnerships, with the goal of gathering large quantities of data that ‘reflect human society.’ The company also said it’s working on making private datasets — such as with data that organizations or companies don’t want to share with others — that can also be used to train AI.”
The AI we see before us, is the direct result of a technology industry, fusing with the military and with intelligence agencies, and their unending quest to find, seize, steal and store our data, and their never-ending greed and hunger to use it to manipulate us, change our behavior, surveil us, imprison us, and so on and so forth. This did not emerge from the needs of users, which should be no surprise considering the absolute bullshit we are being given vs. what the actual entities that AI and technological development are for, have by the metric fuckton. We are getting royalty screwed.
Despite the fact that the CIA is definitely fudging the numbers for marketing on its AI investments, CIA has definitely been very busy little bees in promoting and supporting this agenda, and everything that led up to this AI moment, and the cutting edge of AI for military and intelligence. I went through the portfolio recently just to offer some insight on what some of the focal areas are, so I’ll give a short overview and then we are going to talk about which specific startups — six of them — and technology applications are most concerning here, particularly in the area of warfare.
One theme of the CIA portfolio — the collection of startups it has funded — is the idea of uniting intelligence data — sensitive data that is already inside and streaming to intelligence and other agencies and entities in proprietary infrastructure — with real-time and open data; this has something to do with how data is often siloed, with the CIA having its own propriety data stores, but needing to collect and store and analyze data stores that don’t reside in its own infrastructure, or to capture it for such, for example, in the personal CIA and intelligence agency cloud that Amazon specifically built to meet the CIA’s requirements, in a deal announced in 2014, about in the middle of the web 2.0 bubble, and ushered in by Jassey, once the head of AWS and now the CEO of Amazon itself. How to integrate existing data structures and frameworks with the many sources of data that the CIA is accessing is of course a central concern for them.
Big theme across these companies is bringing “consumer” technology like AR/VR into military contexts. In fact, the entire concept of In-Q-Tel is to benefit from the accelerated development of the consumer market and bring those competencies and data sets and so on, to bear on the CIA’s own information needs. For this reason you see In-Q-Tel investing in a number of startups in the retail and consumer space. In the portfolio we’ve got AI that does digital modeling of houses, that processes insurance, that serves financial institutions. Big focus on retail, for example, a startup that tracks complex physical movement and behavior at retail stores and translates that into, well, actionable intelligence. We see a smattering of database companies — this is a core infrastructure layer and the CIA has invested heavily in these elemental building bricks across the last bubble and into today. We see a big focus on threat detection, security, etc., after all, the CIA has to make sure that not only it but all its constituents have the data they have about us, all locked up where no one can access it but them. There are a surprising number of speech-to-text companies, which I imagine has to do with making things like social interactions, phone calls, public and private conversations, shit they hear with their psycho bugs, fucking freaks, etc., more usable by the machines; able to translate a number of speech acts into a data format which is much easier to reason about. The digitization of the physical world, into intelligence data points, is key.
And over and over again, this idea of the “single pane of glass”. This is one of the obsessions of venture capitalists and the tech industry overall. If you work in this industry you have to hear about this bullshit all the fucking time, it never fucking ends, the single pane of glass, the single pane of glass. Zuckerberg is particularly an adherent to this idea, with his long-held belief or rather, action, that multiple internet services, applications, data sources, etc. are provided seamlessly (I.e. Without choice or transparency) to the user, and this of course is a major design principle of the Metaverse as envisioned by Meta. The most rudimentary example of this and the template, being simple, easy dashboards that are taking large amounts of different kinds of data, and presenting them to the user, or CIA agent in this case. I found interesting, several startups focused on edge, remote and low-connectivity regions, for their obvious implications both as technology in-roads into “leapfrog” technology countries and strategies, as well as for use in remote warfare and intelligence operations in low-connectivity regions. Perhaps this tells us something about the character of the new wars they plan on fighting.
I picked out a few startups that I found most concerning in the In-Q-Tel portfolio, and that highlight the *actually* lethal and existential threat of artificial intelligence, that lethal and existential threat being a variety of parties designing AI that will kill people, on purpose; shoving it into battle zones and wars, targeting of people marked for death, using it to power a massively asymmetrical warfare against countries without these technologies, military and intelligence agency surveillance directly resulting in death, etc. AI in these contexts may be used in things like direct targeting, actual lethal drones, autonomous vehicle piloting, “closing the kill chain” (actual military term), battlefield intelligence, and so on. This is about bringing a whole new level of intelligence to the military and to intelligence agencies — this is their wildest dreams coming true. And again, we get a chatbot that sucks.
Again, I know I’ve said this like 30 times, but why are we talking about the specious threats of possible future generalized intelligence that will decide to rise up and overthrow us for the machine empire, fucking relax, why are we talking about this, admittedly quite interesting and compelling, scenario, and not talking about the whole “AI is killing people right now” thing, which is also quite interesting and compelling if you GIVE A FUCK ABOUT WHAT IS ACTUALLY HAPPENING AROUND HERE.
Ah yes. The startup that refers to itself as the “The Mility Metaverse”, not terrifying at all, totally fine, here the focus on integration of AR/VR in the battlefield, as well as war gaming and training, “enables users to rapidly create and explore courses-of-action, distributed across participants in geographically dispersed locations, leveraging open architecture to integrate incumbent and 3rd party systems, AI and datasets for scenario configuration.” The marketing material brags of “infrastructure that enables defence and intelligence allies, domains and functions to create a common operating picture across physical and virtual worlds.” So again we have this idea of “integrating” these things into a single pane of glass. And additionally for field use to “augment decision making”. Hadean is already being used by the British empire, often called “the belly of the beast”, and is London-based, making it somewhat interesting, but not very, that the CIA is funding it. “Hadean's spatial computing infrastructure integrates legacy and novel simulations into a robust and scalable single synthetic environment, enabling next generation multi-domain training and decision support capabilities for the British Army.” Hadean has raised over 70 million buckaroos, in American dollars. What an exchange!
Highly concerning that in addition to the CIA, this startup has funding from Yuga Labs, an a16z startup known for its work on an NFT metaverse and oh yeah, its totally racist Bored Ape NFTs. This is actually a great example of how the venture capital financial instruments work: this is all the same money — a16z — but here it is being funneled through many of their startups, acting as venture firms themselves. This is perhaps one of the littlest understood components of how venture capital moves, in that THE STARTUPS THEMSELVES ARE VENTURE CAPITALISTS; look to a company like Coinbase which actually has its own venture arm; again, this is just an extension of a16z capital. Coinbase’s venture arm has made 30 different investments, including into Yuga Labs, and Arkham Intelligence, whose investors include Sam Altman and Peter Thiel additionally. Oh the tangled webs we weave!
One of the most concerning aspects of Yuga Labs investing in Hadean, being that, Yuga Labs is a racist 4chan troll. From a previous article in this blog: “we actually have the development of fascist startups, something that we see most plainly with Yuga Labs, a web 3.0, a16z-backed company that has as its logo a very slight variation on a well-known symbol of the Nazi SS and had an openly racist NFT collection, a16z driving $450 million to Yuga to spread this fascist content and creating a community of, according to a source, about 10,000-15,000 people.” There has been quite extensive documentation of the racism and 4chan troll origin of Yuga Labs, including a documentary. Yuga money is a16z and that makes Hadean an In-Q-Tel + a16z collab, pretty natural considering that Marc Andreessen is a high-level CIA asset who served on a secret advisory board for, of all sick evil fucks, Pompeo. Good match!!! And a16z and the CIA fund shit together all the time.
One of the other startups that stood out as The Worst to me was Recorded Future, with funding of about 70 million. First of all, what a creepy as fuck name, it really makes you wonder about the type of people who put this together. “By collecting, structuring, and analyzing threat data from all over the internet for the last decade, we have created an Intelligence Graph of the world’s threats.” I’m sorry, what have you been collecting for the past decade without anyone’s consent or knowledge? In intelligence and surveillance oriented startups, you sometimes find the most bold admissions of surveillance, even far beyond the scenarios typically reported on and discussed in media. What is the “threat data”? What data sources did you get it from? How exactly did this whole operation play out, and when did it get started? The website claims their platform tracks 8 billion entities, 10,000 “actors tracked”, 8 million telemetry points, 5 million “company attack surface mappings”. This, like Palantir, is a pretty good example of how technology companies that define the new bubble, have often been around for awhile or come from long lineages; not every AI startup or company is brand new, because many have arrived naturally through the progression at artificial intelligence; this company for example was founded in 2009 and that longevity here is a massive advantage when it comes to data collection over time, and makes it a great incumbent and contender as data stores are turned into intelligence engines at a scale never seen before. A new startup would have to go get itself a massive data store, but these puppies have been putting them together for years, their potentials now unlocked by developments in artificial intelligence.
The site notes, “With more than 1,700 clients across 75 countries, including the governments of 30 countries, over 50% of the Fortune 100 and 40% of the Forbes Global 100, and the largest holdings of interlinked threat data sets, Recorded Future is the world’s largest intelligence company.” Their site features analysis of Iranian and Hamas application infrastructure. This is a very serious company and not at all like the silly, frivolous, inconsequential or inevitably failing, popular conception of startups. Applications of Recorded Future technology include threat and geopolitical intelligence and attack surface intelligence. And despite the obvious and stated military and intelligence use cases, it is used by Amazon. Wells Fargo. Toyota. IBM. And other major transnational corporations. Because it offers these large enemies the platform to “monitor and take action to mitigate threats to your brand”. “Enable your teams to receive brand alerts on external cyber threats and take immediate action. Defend your brand and reduce risk by receiving real-time visibility into dark web mentions, data leakage, infrastructure risks, and brand impersonations.” Calling itself the “world’s largest intelligence company”, Recorded Future serves both sides of the aisle, and that is actually core to how the CIA technological apparatus works, playing off gains in both the consumer and the military space, to yield far faster and greater innovation cycles than they would be able if they weren’t feasting on the consumer market and technological development fueled by consumer usage. So you get this kindof hybrid consumer/corporate/intelligence/military startup in the AI age.
An important point is that one of the main components and transformations of AI, is that this is basically the CIA for hire, the CIA as a platform, the CIA as software and extensible infrastructure, the CIA as a product, the CIA in an increasingly corporate format, able to take advantage of that fusion and form factor in myriad ways. The relationship between the CIA and venture capital, the exchange with Silicon Valley has actually enabled the intelligence apparatus to be replicable, sellable, scalable, sold out mercenary-style, made available as a service, to whoever will bid for it, made available to corporations for a charge, given to “allies”, and extending the data collecting abilities of the CIA itself.
Next we have Falkonry which is in this similar vein of dealing with absolutely gigantic amounts of intelligence data and artificial intelligence being the means that is used to make that useful to, the fucking CIA and various killer arms of the government.
“Military and Intelligence field systems today generate trillions of high speed data points. The volume and velocity is such that it is impossible for human analysts to keep up. Falkonry’s patented AI enables mission success and readiness for defense organizations by reviewing terabytes of operational data… Analysts can use our solutions to perform activities such as exploiting ELINT [electronic intelligence], anticipating maintenance needs, pattern-of-life detection, improving asset readiness, and increasing the speed and quality of decision making.. Falkonry has successfully demonstrated critical insights at scale against trillions of high-speed data points from U.S DoD's field systems. It can analyse large data sets from diverse sources rapidly to reveal actionable intelligence in real-time and to create domain awareness.”
In case you are wondering what “pattern of life detection” is — and there isn’t much information available about it on the open web — but it is basically where they collect and collate and analyze data from different sources “which can be used to create detailed and accurate models of human behavior and activity.”; This post from an AI expert with a military background gives an example as “the geospatial pattern of life detection relies on a range of geospatial sensors, including satellite imagery, aerial photography, ground-based sensors, and social media data. These sensors collect data from a variety of sources, including human activity, weather patterns, and environmental conditions”. So they are getting increasingly complex, detailed, and integrated, multi-faceted information on us, across time, and our “patterns of life”. The ultimate single pane of glass it seems, is having 100% surveillance and data capture of the human being.
Siren is another one that stood out to me, along these similar lines of the new ages of military data exploitation that become available in the age of artificial intelligence. Siren addresses the problem that “cyber is now a major battle space… The merging of classified information with open sources of information is now a critical part of the national security landscape… Siren has the embedded NLP, entity resolution, graph algorithms, geospatial / temporal analysis and security required for National Security.” Here again we see this massive integration of data taking place to make AI work, to bringing together “open sources of information” (our social media, web activity, movement in physical space etc), almost totally provided by venture-funded startups, together with classified information.
We mentioned earlier in the piece, the focus on low-bandwidth areas being a major theme, and addressing that problem here is Immersive Wisdom, which just this month announced a new contract worth $25 million from the U.S Airforce. Immersive Wisdom is designed to work in “Denied, Degraded, Intermittent, and Limited-Bandwidth (DDIL) environments”, “a remote collaborative ops center platform for distributed & disaggregated operations that allows geographically dispersed personnel to effectively work together and act without having to be physically present, even in severely bandwidth-limited environments, using existing desktops and laptops.” Obviously this speaks to the ongoing massively asymmetrical warfare of many kinds being continuously fought against less technologically savvy areas — typically because they have had their homelands destroyed by US imperialism and colonization.
From the press release announcing the contract, this part stands out as germane to the entire function of In-Q-Tel: serving as a bridge between Silicon Valley and the defense, intelligence, and military apparatuses of the United States; In-Q-Tel facilitates adoption of the startups it funds across these planes of military, defense and intelligence. That is perhaps the bigger benefit on offer to startups taking In-Q-Tel money; it is a promise of a large install base and significant contracts, in a word, a future already set in place for these startups, making money off war zones. Thus we see things like this: “This contract is designed to enable the rapid acquisition of the Immersive Wisdom software platform across the USAF and the entire Department of Defense.”
We’re gonna mention one more then wrap this up, I picked Sayari specifically because of its border work. In-Q-Tel has collaborated with Silicon Valley on the human rights atrocities at the US/Mexico border for some time; most impactfully through its early investment in Anduril, the monster Silicon Valley weapons startup, which has been collaborating with Israeli weapons company Elbit at the border. It is of course important to mention again that Anduril has become a massive military AI player, with a host of AI products for battlefield surveillance and intelligence, AI piloting, drone swarms, autonomous vehicles, “counter intrusion”, and so on. And the CIA was on the ground floor of that too.
Borders are as ever foundational to the imperial and colonial state apparatus, and Sayari is another example of the investment into this space: “Sayari connects key commercial networks and relationships sourced from global public data. By providing instant access to hard-to-find connections, Sayari helps investigative teams eliminate information gaps and expose hidden risks. With visibility into opaque markets and a specific focus on strategic rivals, rogue regime governments and permissive emerging markets exploited by transnational organized criminal networks, Sayari empowers organizations with worldwide transparency at scale, paving the way for safer global commerce.” They have also done work specific to anti-China and Russia action in these days of extremely high tensions.
Welp, that was a great tour of horrors. Again, I demand that we start talking about THIS in relation to AI.
Perhaps one of the things you were thinking as you read this was, damn, you bitches are building all this and we get a lousy chatbot that’s like a slightly better version of Google, but with no sources and a friendly wrapper? Indeed, the computing abilities and potentialities and applications are infinitely beyond what we see at the lowest level, we are essentially the plankton in this situation. Important, yes, necessary even, but it takes a lot of us to feed a whale, and we are definitely getting eaten. We exist in this system primarily as sources of data for the ACTUAL artificial intelligence apparatus.
For more information on how the CIA functions as a venture capitalist arm, here’s an earlier post where I talk more about this fusion and its implications and functions.