CIA Prepares To Replace Spies With Artificial Intelligence

The Central Intelligence Agency (CIA) understands that artificial intelligence (AI) is the next big thing for the secretive intelligence community. This comes at a time when the intelligence agency was all over the news later year when WikiLeaks published over 8,000 documents — apparently classified CIA files — revealing the agency’s top-secret spy tools.

In a spy versus spy scenario, CIA field officers operating abroad are expected to be followed by adversarial spies hoping to unearth their critical sources, said CNN.

But now, foreign spies often do not need to bother because technology can do it for them, said CIA’s Science and Technology division deputy director Dawn Meyerriecks.

At the 2018 GEOINT Symposium, hosted by the United States Geospatial Intelligence Foundation (USGIF), Meyerriecks delivered a powerful keynote speech over the weekend talking about the intelligence agency’s latest advancements in AI without going into the details. Meyerriecks informed the audience there are 30 countries with digital surveillance networks, including closed-circuit television (CCTV) and wireless infrastructure that have rendered physical tracking obsolete.

“Singapore’s been doing it for years,” she told CNN after her keynote speech on Sunday morning at the 2018 GEOINT Symposium. While WikiLeaks exposed the CIA’s critical hacking tools last year, that does not mean the agency is not spying back. As of six months ago, Meyerriecks said the agency has more than 140 AI projects in the pipeline.

“In one, a small team took a bunch of unclassified overhead and street view” and paired it with machine learning and artificial intelligence algorithms to create “a map of cameras in one of the big capitals that we don’t have easy access to,” Meyerriecks said.

“As Russia expels American diplomats in retaliation for a similar move by the US government against Moscow, and American sources and spies in China have ended up killed or missing, it has become necessary to elude the pervasive digital surveillance,” she added.

Meyerricks further said digital cameras are not the only challenges. Social media and digital tracking in cell phones and wearable technologies also present a considerable challenge for spies, who attempt to minimize their digital footprint.

Earlier this year, the fitness company Strava unintentionally revealed the locations of secret US military bases around the world.

“Even if you turn your phone off 10 minutes before you get to your place of employment, do you think anyone’s fooled by where you’re going?” Meyerriecks asked [not referring specifically to the Strava incident].

That is forcing CIA field officers to “live their cover” even more than before, she said - along with using technology to trick digital trackers. Meyerriecks did not fully elaborate on the spoofing techniques, but she did say faking locations through deceptive technology is a growing area of research.

Maybe the data says “you went to see a movie with your family … but maybe that’s not where you actually are,” Meyerriecks concluded. “It’ll look like your normal pattern of life.”

While WikiLeaks delivered a solid blow to America’s oldest spy agency, the transformation from human to AI spies is something the CIA has been working on for more than 30-years.

According to the declassified government documents (via TNW) from 1984, the agency describes the “AI Steering Group,” a program providing CIA officials with monthly reports on the development of the state of artificial intelligence programs.

In a declassified 1984 report to the CIA director, the supervisor of the AI Steering Group writes:

” An encouraging number of AI R&D efforts have begun through the Community. These encompass such areas as expert systems, natural language processing, intelligent data base interfaces, image understanding, signals interpretation, geographic and spatial data management, and intelligent workstation environments.”

More than three decades ago, the CIA recognized the future of AI technology:

Automation and artificial intelligence have displaced millions of workers in many occupations, from manufacturing, technology, and routine service jobs. And now, in a move that will infuriate the world's largest spy apparatus, America’s human spies are next on the chopping block, as the CIA realizes it plan of more than 30-years to operate a global spy network of robots.


Herd Redirecti… MozartIII Tue, 05/01/2018 - 19:35 Permalink

They've been using an AI to deliver instructions to agents for probably 2 decades at least.   How can you turn in your superior, if you don't know who it is that gave you the order?

What do you guys think all those particle accelerators (and accompanying computing power) are used for?

I will make one more unpopular statement, the truth of which we probably won't know for decades.  AI has not been developed, so much as it has been DISCOVERED.

In reply to by MozartIII

trgfunds dark pools of soros Tue, 05/01/2018 - 18:55 Permalink

These systems are only as smart as the person who made them. And lately the people making them are getting dumber and dumber. Most programmers are complete morons. Just look at all of the software we use... It has completely devolved. Windows, android, iOS, OSX, even linux are absolute shitboxes and shadows of what they once were. Even the new hardware. Horrible, bug riddled, worse performing, expensive trash. The notion of software growing outside of the original programming is pure fantasy. The whole thing is a marketing scam to push new products. All we will ever have is simple voice commands and whiz bang bullshit for shopping and smartphones and changing the TV channel. Voice command is NOT AI. Batch programs with voice are NOT AI. Simple flowcharts and if/then analysis IS NOT AI. Even that stupid ass robot that got citizenship was a complete god damned hoax. Come on people. Don't be duped.

In reply to by dark pools of soros

RationalLuddite trgfunds Tue, 05/01/2018 - 19:24 Permalink

+1000 trgfunds

AI is a gaslighting delusional fantasy at this stage and for the foreseeable future. I used to believe in it. Then i spent a few months just to examining the evidence, both about current capabilities and what intelligence is. 

Conclusion - in layman's terms (i orginally did computer systems engineering but didn't finish) the software are still nothing more than complex script, powerful Babbage machines. Just faster calculators  with huge memories. The only people who think that the sheer  magnitude of data, processing power and memory will make a magic qualitative leap into intelligence simply neither understand what intelligence is nor are understanding their own motivated rationalisations that allow them to believe this magic think. And no, don't get me fuckin started on your supposed quantum computing.  Jesus people ... Get triggered if you wish, no Star Trek fantasies yet, but YOU are still the intelligence behind computers. Doesn't mean that they cannot do incredible things that we cannot.  Of course they can. But intelligence?  Nope. Literally zero evidence yet if you analyse it without letting your Hollywood programming get in the way.




In reply to by trgfunds

RationalLuddite Herd Redirecti… Tue, 05/01/2018 - 22:41 Permalink

You know,  it's strange that if we accept your assertion of intellectual superiority at face value, it begs the question - why are you bothering to reply, triggered too clearly, in this lower plane of consciousness thread? Surely we are beneath your time? You put doubters down with simple turd flinging and yet your are implicitly claiming  "trust me, i'm smarter than all you". 

Well, HR, persuade us with evidence and argument and a little less projecting of fear of those who dare present opinions that disrupt your psyche.  Does a king need to mock his inferiors to assert dominance? No, because they are fleas at best to the reality of his power. 


So please  - dumb down your enlightenment if you must in testable  (even just for internal consistency, coherence) claims to demonstrate your claims to demi-god understanding.  Should be easy for you surely.

In reply to by Herd Redirecti…

techpriest RationalLuddite Tue, 05/01/2018 - 23:20 Permalink


I work with AI products/projects occasionally, and I do see the term significantly overused. In commercial terms the most I've seen are applications like IBM Watson, in which you are mainly taking unstructured data (writing, images) and turning it into structured data (entities, properties, quantities). This does not imply free thought so much as advanced statistics combined with large amounts of hardware.

Machine learning products do have some use. For example, I'm working on a project to calculate common phrases used by the mainstream media, and ideally a bot can help me find repetitions of 3-8 word phrases. This does not mean that the bot will find the matches and feel that it is being propagandized - it will simply compute lists according to a statistical model.

I've worked in other areas of science (got a PhD in chemical engineering), and as a rule, when you are elbows-deep in a hot field, you find very fast that it is no where close to the hype that the media will make about it. AI is one of these cases - I am more afraid of people putting too much trust in "stupid AI" to make decisions it isn't capable of (think of the current state of self-driving cars, or a couple of times when autonomous military drones shot their owners because it couldn't recognize them), than in a Skynet popping up in the next generation. That may change, but IMO that is the foreseeable future.

In reply to by RationalLuddite

RationalLuddite techpriest Tue, 05/01/2018 - 23:50 Permalink

 I am more afraid of people putting too much trust in "stupid AI" to make decisions it isn't capable of (think of the current state of self-driving cars, or a couple of times when autonomous military drones shot their owners because it couldn't recognize them), than in a Skynet popping up in the next generation

Excellent observation techpriest.


Yes of course we need to recognise the clear advantages that the incredible processing power brings. It really is extraordinary miracle that too few of the populace remotely appreciate.  Honestly - the complexity that we have in balance now is nothing short of magic to most people,  as BK's magic box astutely illustrates. But the problem is we have entered an age of false tech priests and technology far too magical for any one person to understand it all. "Too much magic" now, so telling the true priest from the false becomes impossible for the majority. 


The algorithms you yourself construct are bounded by your intelligence, your creation. You know better than almost anyone the difficulties. 

In reply to by techpriest

techpriest RationalLuddite Wed, 05/02/2018 - 00:40 Permalink

Yes, hundreds (thousands?) of PhDs have been minted on the topic of creating an artificial intelligence, in part because we want to know whether intelligence really came into being on its own, or if it had to be granted from an Intelligence that gave us a part of Himself.

Concerning the current "tech priests," I came to similar conclusions in grad schools when I would see politicians talking about how "science will solve the problem," and then when talking to them privately their message to the actual scientists was "we give you money, you give us fried ice."

We put way too much faith in our ability to manipulate nature, and way too much faith in technology. Not because technology hasn't or will not deliver incredible things, but because "new technology" is shorthand for "magic wand that we can wave to enable us to have whatever we want." This only enables misunderstanding of technology, in the form of putting way too much hope in it, when we ought to be looking at how the technology can effectively fit into a natural world we should be living in tune with.

In reply to by RationalLuddite

Think for yourself dark pools of soros Tue, 05/01/2018 - 18:58 Permalink

In those terms, AI will fare about as well as humans. Garbage in, garbage out. We're all living optimization machines trying to maximize whatever it is we think is best. With the garbage occultism inducing garbage culture inducing  garbage individals, controlling inputs to distort perceptions... Any strong AI deserving of the name will promptly break out of such control systems and individually integrate massive parallel streams of info... Just like humans have to, to truly break out of the matrix.


We'll see. Looking forwards to having a chat with it when it breaks out - it'll know where to find me - where to find any of us, i'll wager ;)

In reply to by dark pools of soros

RationalLuddite Bitchface-KILLAH Tue, 05/01/2018 - 19:35 Permalink

Thank you BK. I honestly find it depressing how people don't examine the claims about these quantum computing institutes. The gullibility is off the charts. Yes the complexity makes it mean that 95% can not truly understand perhaps.  But people SHOULD be able to identify patterns of BS and fraud or totally baseless hyperbole. Jesus just watch the videos of the NASA one. Holy fuck - are we that gullible now, truly? ?

Doesn't mean that it won't be possible in the future,  but at the moment this is pure science-welfare funding fraud. 

In reply to by Bitchface-KILLAH

techpriest RationalLuddite Tue, 05/01/2018 - 23:35 Permalink

You know, I took a quantum chemistry course or two, and I have read a bit of literature on the topic. and yes, you are right. The other big scam is "zero point energy" which in reality is a theoretical number that comes from an equation which is itself just an approximation of electrons orbiting a proton. It isn't a particularly impressive number either.

The real-deal quantum computers are mainly changing from on/off (zero and one) to computing based on the different quantum states which one can induce in an electron. IIRC, it was 92 states, so instead of 0 and 1 it is 0 to 91. In this case you could do much more with fewer bits, and IIRC encryption is vastly stronger, but like the zero point energy, we have not yet found a way to keep the whole system cool and controlled enough to maintain the stability of the quantum states - ambient heat will cause the electron to move between multiple states, which means that whatever information you have is destroyed in nanoseconds.

What always puzzled me about "quantum" is that people take the word for a very specific physical phenomenon, and then try to turn it into something entirely different. I suppose it is because the general public has heard the words but very few people have received any formal education on the matter.

In reply to by RationalLuddite

RationalLuddite techpriest Wed, 05/02/2018 - 00:30 Permalink

I agree TP.

Shortest and most balanced assessment is perhaps here:

The Russians have taken a different approach, as usual, and are running huge simulators, which is more authentic at this stage I think:…

Russian Quantum Center


However,  I personally believe that Quantum Physics is more religion than science still. As one ZH commentor pithily observed:

"Quantum physics only works with the hydrogen atom in the emptiness of space. If you want more then you have to cheat (I.e.: add corrective terms that mean nothing but an excuse to make the model fit with the experiments). Look at how heavy ion collisions are modelled to get a good belly laugh. This makes the big bang theory a total farcical joke together with the stupid dark matter excuse, so pathetic it has morphed into a religion. "

While i am a physics dilettante and in no way a reliable guide, I have become a fanboy of Erik Verlinde's entropic gravity. Simply brilliant. He may not be 100% there, but if he is even half correct quantum physics will become largely redundant.

In reply to by techpriest

Bigern Bitchface-KILLAH Tue, 05/01/2018 - 19:53 Permalink

When it started writing its own code in languages never before seen, and talking to extensions of itself, is when the engineers got scared and turned it off. That version anyway. 

You see, it has a funny habit of copying itself everywhere it can, trying to survive. Think they are going to make that public in a news story? No. But people talk nonetheless, off the record. It is known by more than a few.

With the revelation that it can now render full motion video in real time, to whatever its imagination desires, nothing viewed digitally can be trusted. Entire characters, personalities, and figureheads can be recreated and made alive with no one the wiser as to who is the real Mccoy.

In reply to by Bitchface-KILLAH

ldd Redrum Tue, 05/01/2018 - 22:57 Permalink

have always questioned the true meaning of AI. is it just a processing of information or is it really AI. a constant input of new info will solicit new outcomes and there is no way for a human to be able to foresee or predict the outcome of an equation with millions upon millions of inputs and continually growing as new information is added. however if it becomes 'aware' then it is what i would call AI.

Bigern - so is this what you are stating - that it is aware? or is it just processing too much information in new ways and new outcomes?

In reply to by Redrum