The key takeaway is that setting an explicit value function to an AGI entity is a good way to destroy the world due to poor AI alignment. We are more likely to not destroy by creating an AI whose goals is to "do want humans what it to do", but in a way that it does not know before hand what it is that humans want, and it has to learn from them. This approach appears to be known as reward modeling.
Some other cool ideas:
- a big thing that is missing for AGI in the 2010's is some kind of more hierarchical representation of the continuous input data of the world, e.g.:
- intelligence is hierarchical
- we can group continuous things into higher objects, e.g. all these pixels I'm seeing in front of me are a computer. So I treat all of them as a single object in my mind.
- game theory can be seen as part of artificial intelligence that deals with scenarios where multiple intelligent agents are involved
- probability plays a crucial role in our everyday living, even though we don't think too much about it every explicitly. He gives a very good example of the cost/risk tradeoffs of planning to the airport to catch a plane. E.g.:
- should you leave 2 days in advance to be sure you'll get there?
- should you pay an armed escort to make sure you are not attacked in the way?
- economy, and notably the study of the utility, is intrinsically linked to AI alignment
Some criticisms:
Ciro Santilli invented this term, derived from "hardware in the loop" to refer to simulations in which both the brain and the body and physical world of organism models are modelled.
E.g. just imagine running:
- NeuroMechFly
- a Drosophila connectome simulation
Likely implies whole brain emulation and therefore AGI.
It's like the bootloader stage of biology! It's weird and magic and important: Section "Molecular biology feels like systems programming".
CIA 2010 covert communication websites by 
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01This article is about covert agent communication channel websites used by the CIA in many countries from the late 2000s until the early 2010s, when they were uncovered by counter intelligence of the targeted countries circa 2011-2013. This discovery led to the imprisonment and execution of several assets in Iran and China, and subsequent shutdown of the channel.
The existence of such websites was first reported in November 2018 by Yahoo News: www.yahoo.com/video/cias-communications-suffered-catastrophic-compromise-started-iran-090018710.html.
Previous whispers had been heard in 2017 but without clear mention of websites: www.nytimes.com/2017/05/20/world/asia/china-cia-spies-espionage.html:
Some were convinced that a mole within the C.I.A. had betrayed the United States. Others believed that the Chinese had hacked the covert system the C.I.A. used to communicate with its foreign sources. Years later, that debate remains unresolved.[...]From the final weeks of 2010 through the end of 2012, [...] the Chinese killed at least a dozen of the C.I.A.’s sources. [...] One was shot in front of his colleagues in the courtyard of a government building — a message to others who might have been working for the C.I.A.
Most notably, starting in 2008, CIA contractor John Reidy started raising concerns about the security of the communication systems used, but he was silenced and ignored, leading to catastrophe.[ref][ref]
Then in September 2022 a few specific websites were finally reported by Reuters: www.reuters.com/investigates/special-report/usa-spies-iran/, henceforth known only as "the Reuters article" in this article.
Banner of the Reuters article
. Source. 
Inspecting the Reuters article HTML source code
. Source. The Reuters article only gave one URL explicitly: iraniangoals.com. But most others could be found by inspecting the HTML of the screenshots provided, except for the Carson website.Ciro Santilli heard about the 2018 article at around 2020 while studying for his China campaign because the websites had been used to take down the Chinese CIA network in China. He even asked on Quora: www.quora.com/What-were-some-examples-of-the-websites-that-the-CIA-used-around-2010-as-a-communication-mechanism-for-its-spies-in-China-and-Iran-but-were-later-found-and-used-to-take-down-their-spy-networks but there were no publicly known domains at the time to serve as a starting point. Chris, Electrical Engineer and former Avionics Tech in the US Navy, even replied suggesting that obviously the CIA is so competent that it would never ever have its sites leaked like that:
Seriously a dumb question.
So when Ciro Santilli heard about the 2022 article almost a year after publication, and being a half-arsed web developer himself, he knew he had to try and find some of the domains himself using the newly available information! It was an irresistible real-life capture the flag. The thing is, everyone who has ever developed a website knows that its attack surface is about the size of Texas, and the potential for fingerprinting is off the charts with so many bits and pieces sticking out. Chris, get fucked.
In particular, it is fun to have such a clear and visible to anyone examples of the USA spying on its own allies in the form of Wayback Machine archives.
Given that it was reported that there were "more than 350" such websites, it would be really cool if we could uncover more of those websites ourselves beyond the 9 domains reported by Reuters!
This article documents the list of extremely likely candidates Ciro has found so far, mostly using:more details on methods also follow. It is still far from the 885 websites reported by citizenlabs, so there must be key techniques missing. But the fact that there are no Google Search hits for the domains or IPs (except in bulk e.g. in expired domain trackers) indicates that these might not have been previously clearly publicly disclosed.
- rudimentary IP range search on viewdns.info starting from the websites reported by Reuters
- heuristic search for keywords in domains of the 2013 DNS Census plus Wayback Machine CDX scanning
If anyone can find others, or has better techniques: Section "How to contact Ciro Santilli". The techniques used so far have been very heuristic, and that added to the limited amount of data makes it almost certain that several IP ranges have been missed. There are two types of contributions that would be possible:Perhaps the current heuristically obtained data can serve as a good starting for a more data-oriented search that will eventually find a valuable fingerprint which brings the entire network out.
- finding new IP ranges: harder more exiting, and potentially requires more intelligence
- better IP to domain name databases to fill in known gaps in existing IP ranges
Disclaimer: the network fell in 2013, followed by fully public disclosures in 2018 and 2022, so we believe it is now more than safe for the public to know what can still be uncovered about the events that took place. The main author's political bias is strongly pro-democracy and anti-dictatorship.
May this list serve as a tribute to those who spent their days making, using, and uncovering these websites under the shadows.
If you want to go into one of the best OSINT CTFs of your life, stop reading now and see how many Web Archives you can find starting only from the Reuters article as Ciro did. Some guidelines:
- there was no ultra-clean fingerprint found yet. Some intuitive and somewhat guessy data analysis was needed. But when you clean the data correctly and make good guesses, many hits follow, it feels so good
- nothing was paid for data. But using cybercafe Wifi's for a few extra IPs may help.
viewdns.info
. Source. activegameinfo.com domain to IP
viewdns.info
. Source. aroundthemiddleeast.com IP to domain
DNS Census 2013 website
. Source. This source provided valuable historical domain to IP data. It was likely extracted with an illegal botnet. Data excerpt from the CSVs:amazon.com,2012-02-01T21:33:36,72.21.194.1
amazon.com,2012-02-01T21:33:36,72.21.211.176
amazon.com,2013-10-02T19:03:39,72.21.194.212
amazon.com,2013-10-02T19:03:39,72.21.215.232
amazon.com.au,2012-02-10T08:03:38,207.171.166.22
amazon.com.au,2012-02-10T08:03:38,72.21.206.80
google.com,2012-01-28T05:33:40,74.125.159.103
google.com,2012-01-28T05:33:40,74.125.159.104
google.com,2013-10-02T19:02:35,74.125.239.41
google.com,2013-10-02T19:02:35,74.125.239.46
The four communication mechanisms used by the CIA websites
. Java Applets, Adobe Flash, JavaScript and HTTPS
Expired domain names by day 2011
. Source. The scraping of expired domain trackers to Github was one of the positive outcomes of this project.Compromised Comms by Darknet Diaries (2023)
Source. It was the YouTube suggestion for this video that made Ciro Santilli aware of the Reuters article almost one year after its publication, which kickstarted his research on the topic.
Full podcast transcript: darknetdiaries.com/transcript/75/
Related to Leela Zero, a Go engine
One of the most powerful chess engine as of 2023: computer chess competition.
CLI program implementing Universal Chess Interface: www.reddit.com/r/ComputerChess/comments/b6rdez/commandline_options_for_stockfish/
How to actually play against it: chess.stackexchange.com/questions/4353/how-to-install-stockfish-on-ubuntu So hard!
The user friendly Chess UI! Exactly what you would expect from a GNOME Project package. But also packs some punch via the Universal Chess Interface, e.g. Stockfish just works.
Both chess engine and a CLI chess UI. As an engine it is likely irrelevant compared to Stockfish as of 2020. TODO: does the UI support Universal Chess Interface?
Cool project history though. Started before the GNU Project itself, and became one of the first packages.
Advanced. Not beginner friendly, very clunky.
To be fair, this is one of the least worse ones.
The cool thing about this notation is that is showed to Ciro Santilli that there is more state to a chess game than just the board itself! Notably:plus some other boring draw rules counters.
- whose move it is next
- castling availability
- en passant availability
There are unlisted articles, also show them or only show them.