Liberation Travel Hacks 05/2026

Dear friends,

after the great Bratislava event Bitcoin Je Retro I moved on to Japan to visit my friends in Hokkaido and to explore the rest of Japan for a few days. The original plan was to cross all of Japan by bicycle, but given the lack of time I'm pushing it to next year (I need at least two continuous months for that).

First Japanese surprise — Japan is not the “bleeding-edge” technology country from the year 2100 that many people might imagine (you'd get that feeling more in South Korea or in big Chinese cities). It's “functional retro” — many technologies here are 20–30 years old, but since they're maintained, there's no reason to change them. It also means that neither Chinese nor European dystopia has arrived here yet. And we get to the second Japanese surprise — I pay for almost everything in cash, no card, no app, no QR code. Japan is a cash-based society. At the same time, there are no state-enforced cash limits here (unlike in the EU). Which I see as more than positive, because the state clearly doesn't want to mass-surveil the financial transactions of its citizens (as is the case in the EU, where it keeps getting worse). And it doesn't invent stupid pretexts that it's for the “well-being and safety of all citizens”. There's no Chat Control here, no MiCA, online anonymity is the norm (and they don't plan to change it). Of course, they have some stricter laws than in the EU (obscenity ban, hard drug laws). None of that changes the fact that Japan has higher economic / financial freedom than the EU and should serve as inspiration for European comrades to deregulate.

One of the most amazing things about traveling is that it allows me to understand that the vast majority of laws and regulations make no sense in a civilized society and dramatically restrict our freedom.

In Uruguay, all drugs are legal for personal use and society has dealt with it (banning them would have the opposite effect). In Paraguay there's not only no mandatory health insurance, but also no mandatory car insurance — everything is voluntary, and society not only functions without problems but keeps growing (Moody's lowers Slovakia's rating every year, and raises Paraguay's).

You don't need to travel a lot. Open your eyes, look around you — and you'll see how many things can work freely and voluntarily.

Pavol Lupták, 3rd of May 2026, Tokyo, Japan

Germany

Thanks to an amendment to the German compulsory military service law, starting 1 January 2026 German men aged 17–45 must obtain permission from the German armed forces — even outside cases of tension or defense — if they want to leave Germany for more than 3 months.

The legislation itself. Staatenlos's reaction.

If there is suspicion that they won't return within 3 months, a German passport simply won't be issued to them.

Until now this only applied during military threat; now it's permanent. Permission to travel has a long tradition in Germany (GDR). Not only in times of threat, but now also in peacetime, the state has control over your body — deciding where a citizen may or may not be located. My German friends are in full panic mode. My Uruguayan colleagues report an enormous number of applications for Uruguayan residency from German citizens. And the rest of us are waiting to see when "travel permission" will arrive from Germany to all the other EU countries.

EU

It sounds harsh, but as far as internet control goes, the EU is following the Russian playbook.

These provisions prove that the EU has decided to take unprecedented measures to restrict freedom of speech on the European internet. In fact, it could be called an organized legal war against freedom of speech online.

A more dystopian name than "Protect what matters” really can't be invented, in my view. I think George Orwell would praise European officials for this. In 2026 and 2027, parliamentary, regional, or presidential elections await

Juraj Bednár: EU age control: a Trojan horse for digital ID.

What AI thinks about European age control (Opus 4.7):

The effectiveness of Age Control legislation in reducing child abuse is, based on available data, close to zero, possibly marginally positive in terms of porn exposure for the least motivated children. At the same time:

It addresses the wrong problem. 93% of abuse happens offline, beyond the reach of any digital controls.
It addresses the wrong minority of the online problem. CSAM does not circulate on the platforms that AC will affect.
It is trivially circumventable. Empirical data from the UK (+6,430% VPN) and Australia prove this in real time.
It creates infrastructure with permanent costs to privacy and to the architectural freedom of the internet.
It is politically cheap. It provides strong emotional rhetoric (“we protect children”) without the costs that real measures would carry — resources for social work, investigative police on the darknet, parental education, legislation against intra‑family abuse.

Real reduction of child abuse would require: more resources for CPS and child psychologists, better detection protocols in schools and healthcare, programs for at‑risk families (single‑parent households with step‑partners), investigative capacity targeted at darknet CSAM distribution networks, a “security by design” approach for online platforms, and educating children about boundaries and communication with adults.

None of that is a priority in the current AC package. That alone is the strongest argument that the declared goal and the actual goal diverge.

Full analysis with sources.

A 2024 study analyzing content removed by platforms in Germany, France, and Sweden found that 87 to 99 percent of removed content was legal speech. Platforms simply prefer to over‑remove rather than risk a fine of up to €50 million under the German NetzDG, or up to six percent of global turnover under the DSA.

A US Judiciary Committee report from February 2026, based on documents the committee compelled from platforms, describes in detail how the European Commission, over ten years, systematically pressured platforms to change content moderation rules globally — i.e. including content of US citizens in the US. Internal TikTok documents confirm that the company censored legal speech and even truthful information worldwide in order to comply with DSA requirements.

In Germany, the situation has descended into dystopian absurdity. As Yascha Mounk documented, the German Interior Minister reported dozens of citizens to the police for critical comments addressed to her. A journalist who, in protest, posted a satirical meme received a seven‑month suspended sentence. Human Rights Watch labelled the German NetzDG law “fundamentally flawed” and warned that it turns private companies into “overzealous censors”.

UK

The state "knows best" what "is good" for your health: If you approve of the new British smoking ban because it doesn't affect you (you don't smoke), or you claim "smoking is bad, therefore it must be banned across the board", then you completely miss the essence of individual freedom.

In that case you logically have to approve banning desserts, oversweetened drinks, and mandatory exercise for all citizens, since these things — like cigarettes — shorten people's lives, and the resulting cardiovascular diseases and diabetes kill hundreds of millions of people a year. I never smoked, but how a society views smoking is a litmus test of individual freedom for me. And personally, as a non‑smoker, I'd rather live in the free Balkans, where everyone smokes, than in the UK, where people will be criminalized for smoking.

In the United Kingdom a smoking ban for persons born after 2008 was approved.

Georgia

Samuela Davidová and Giorgi Tushurashvili – Crypto‑friendliness in Georgia 🇬🇪 in 2026. Crypto vs. banks.

New rules for LLC companies in Georgia, work permits, and remote business in 2026.

7 reasons why you want a TBC Concept account (even if you already have an account at Bank of Georgia).

All TBC Concept 360 customers get free 3 GB of Airalo roaming.

Opening a TBC bank account (Georgia).

Paraguay

A new service for our clients – paying invoices from Paraguayan Personal with any crypto – without needing a Paraguayan bank account, a Paraguayan payment card, or a Paraguayan IP address. Pay Personal Paraguay mobile operator bills via Bitcoin, stablecoins, or Monero/Zcash. Personal is the best Paraguayan mobile operator with worldwide data roaming, 80 GB monthly for 200K PYG (€26).

Paraguay this month approved the “Investor Pass” program via SUACE — a fast track to permanent residence and tax residency for foreign investors. Unlike the original SUACE investment program ($70,000 + business plan + 5 employees), here you really only need money and can apply for permanent residence directly. It's enough to:

make an investment in real estate, at least $200,000

make an investment in tourism, at least $150,000

buy securities of at least $200,000

According to our lawyers, it's enough to sign a purchase‑sale contract and pay at least 30% of the total amount (and the rest gradually). We already have two interested clients, so we will be testing the Investor Pass in detail over the coming months.

Paraguay's unfortunate crypto legislation is still in force (and we're waiting for it not to be). By the way, since the start of this year there have been in France 41 crypto kidnappings carried out (!), one of the main reasons being that French tax‑office employees sold information about crypto taxpayers to terrorists. And Paraguay is a more corrupt country than France. If, as Paraguayan tax residents, you haven't yet signed the petition, now is the time! (it's already been signed by almost 8,000 people).

Backpacking through Paraguay (2026): Best places, routes.

Paraguay is currently perhaps one of the most underrated countries in the world. Why?

France

Despite the fact that in France a hapless crypto person who declared their crypto to the French state is kidnapped every few days, France is taking inspiration from Paraguay and is about to pass similar surveillance legislation. France is adopting a law that requires anyone holding more than €5,000 in a crypto wallet to report it to the tax authorities. This is the same government whose tax‑office employee sold the data of people who declared their crypto to criminals, who then kidnapped their families and cut off their fingers. French officials will have blood on their hands.

Turkey

I love Turkey, but I like the local government all the less for it. You should know that Turkey is becoming a new tax haven and is introducing a 20‑year tax exemption for foreigners. Erdoğan has just dropped an enticing offer: foreigners who haven't been tax residents in Turkey for the last 3 years can now settle there and, for 20 years, pay no taxes on their income or capital gains from abroad. Only income earned in Turkey will be taxed. The goal is apparently to grab a large slice of Dubai's golden clientele and to attract European talent fleeing excessive regulation and stifling taxation.

Russia

This is not just criticism of the current state in Russia, but a reminder that thanks to the European digital ID we are heading toward a similar scenario.

How military conscription happens in Russia

Gosuslugi (the Russian digital ID system) has become a central hub without which life in Russia is almost impossible. Through it, people obtain documents, schedule doctor appointments, and more recently, receive electronic military summons. If a citizen does not confirm receipt, their rights are automatically restricted (leaving the country, selling property, driving a car) — which is a direct implementation of digitally “switching off” an individual from the system.

This represents one of the most radical examples of a state transforming into a digital control mechanism in modern history. What was once a portal for paying parking fees or booking passport appointments has become a digital cage.

Before this law (2023), a military summons in Russia had to be delivered in person and signed for. People would simply avoid opening the door or live at different addresses. Now, a summons is considered delivered the moment it appears in your personal Gosuslugi account — or, if you do not have an active account, 7 days after it is entered into the “Unified Register of Conscripts.”

The travel ban activates immediately upon delivery of the summons — before the 20-day window begins. If a citizen then fails to report to the military office within 20 days, the following additional restrictions are triggered:

Driving ban: The driver’s license becomes invalid in the traffic police database.
Ban on buying/selling real estate: The land registry (Rosreestr) blocks any transactions. You cannot sell property to leave the country.
Credit ban: Banks automatically see that the applicant is subject to conscription restrictions and loan access is blocked.
Business restriction: You cannot register a company or work as a freelancer (self-employed).

In Russia, it is nearly impossible to function without a Gosuslugi account. You use it to enroll your child in school or kindergarten. You receive QR codes through it (as seen during the pandemic).

When people began attempting to delete their accounts in March 2023 ahead of the spring conscription drive, the authorities disabled the “delete account” option on the website. This happened proactively on March 31, 2023 — the same day the Defence Ministry announced electronic summonses — rather than in response to a mass exodus already underway.

The system is integrated with a network of over 200,000 cameras in Moscow equipped with facial recognition. In documented cases, the flagging has occurred specifically when a conscript contests his draft order in court — at which point the enlistment office enters him into the system as an alleged evader, triggering a facial recognition alert and enabling police to detain him on the spot. It is not confirmed as a blanket automatic trigger for anyone who has simply received a summons and not yet reported.

The state can effectively cut a person off from the social and economic bloodstream — without the ability to drive, manage money, or use property.

This became reality in Russia faster than almost anywhere else because the state used war as justification to merge all databases (tax, police, medical, and military) into one central system.

Russia no longer uses technology merely as a service for citizens, but as an operational tool of enforcement.

Bitcoin, Monero, Zcash and cryptocurrencies

Juraj Bednár: Lightning as a universal checkout: why I built AnyPay. AnyPay uses Lightning as a source to pay any crypto payment, instantly, privately, without custody.

My friend Tomáš Forgáč is launching in Switzerland the Helva project — a crypto‑to‑fiat lending platform (Swiss quality).

Kashilo — a marketplace where privacy comes first and where Monero is the only payment method. The idea: buy and sell things without creating an “account” (UUID), without disclosing an email address, and without anyone tracking you.

Beware of fake Ledger hardware!

From neobanks to onchain banks.

How to turn crypto into fiat without a paper trail.

Zcash is now available on THORChain.

Privacy

Anonymous prepaid/gift cards for crypto and anonymous eSIMs via the service Stealths.

Freedomia.io is another crypto debit card without identity verification (KYC). The service offers 3DS verification for secure online payments and integration with digital wallets such as Google Wallet.

Nada Nada — No account. No e‑mail. No identity verification. Instant and anonymous VPN, SMS and eSIM services.

Cape is a US privacy‑first mobile carrier (an alternative to Verizon/AT&T/T‑Mobile) at $99/month with unlimited calls, SMS, and 5G data. Main unique features: IMSI identifier rotation every 24 hours (every day you look like a new subscriber), call metadata deleted after one day, last‑mile encrypted SMS, protection against SIM swap and SS7 attacks, anonymous payment without a name via tokenization, and safe roaming in 50+ countries. Sign‑up without personal data, eSIM activation in 2 minutes, full Pixel + GrapheneOS support (and they donate the first month to the GrapheneOS project).

There's more and more talk about the great Q‑day, which marks the future, hypothetical date when quantum computers will be powerful enough to break standard public‑key encryption algorithms (such as RSA) that currently secure global digital communication. The risk that has so far been hypothetical may, in the future, stop being hypothetical — also thanks to the exponential rise of AI. In my view this is a problem that should be addressed by (not only) all cryptocurrency creators. Protecting cryptocurrencies through responsible disclosure of quantum vulnerabilities. An introduction to post‑quantum cryptography.

A researcher broke a 15‑bit ECC key on publicly available quantum hardware, achieving a 512× jump over the previous public demonstration.

The Dutch Court of Audit published an audit report on the fight against money laundering (AML) in banking, which states that “there is a lack of understanding of the effectiveness of the anti‑money‑laundering approach”. The document states that “it is not clear whether increased controls by banks actually contribute to preventing and detecting money laundering”, while highlighting the considerable costs that this measure imposes on banks.

LinkedIn illegally scans your computer. Microsoft is running one of the largest corporate espionage operations in modern history. Every time one of LinkedIn's billion users visits linkedin.com, hidden code scans their computer to determine the installed software, collects the results, and sends them to LinkedIn servers and to third parties — including a US‑Israeli cybersecurity company.

I have to admit, as a tourist in Brazil I really hate the CPF (Brazilian tax number) which you have to enter practically everywhere when ordering anything online in Brazil. I even needed it to buy carnival tickets and tickets to the botanical garden (yes, CPF generators exist). And now the Brazilian government has been completely hacked and sensitive tax information is being sold.

The Bitwarden CLI password manager was compromised in a “supply chain” attack. The page @bitwarden/cli@2026.4.0 contained malicious code after attackers took over GitHub Actions, stole sensitive data, and uploaded a modified version to the npm repository.

AI — commentary

When the technical solution is a breeze, the problem is bureaucracy

Thanks to AI I wake up in the morning with any idea, and by evening it's implemented and tested. That absolutely doesn't apply to the bureaucracy I run into along the way — it's solvable, but takes weeks or months.

The problem I run into daily is no longer technical but bureaucratic:

1. I wrote an application that accepts any link to a booking.com accommodation reservation. Through our API access to wholesale portals it figures out whether we can book it cheaper for the customer, and if yes, it books it directly.
Bureaucratic problem: we need a travel agency resales license to officially offer this service.

2. I wrote a Signal – WhatsApp Bridge for our clients. So they can send WhatsApp messages to anyone without installing WhatsApp (and it maintains the Signal – WhatsApp sessions).
WhatsApp immediately blocked my proxy account, saying I can't offer this kind of service at all (I managed to solve this problem with a special communication obfuscation).

3. I launched an e‑shop for our Liberation.Travel services https://liberation.services— we accept all kinds of cryptocurrencies but no payment cards. The application for a payment gateway is so complicated for an offshore company that I'm not bothering and we're staying 100% crypto only. Actually I'm glad too, because we're not endangering our clients' privacy with fiat services.

4. I wrote an AI bot that posts to all social networks via Signal. LinkedIn requires filling out a bureaucratic application so I can use their API to post to my own corporate LinkedIn groups. X.com wants $200/month before they let me publish anything via their API. I hacked both via SocialCrabs.

Worth noting:

Bureaucracy (not only state‑based but also private) will stand in the way of the massive spread of AI.
Currently 125 bureaucrats in Brussels and thousands of officials across 27 countries are supposed to regulate 4,100+ AI startups (and I'm surprised any exist in the EU at all).

What awaits us in the future:

1. States will introduce "protection" of human labor against AI. I expect they'll introduce subsidies for AI‑replaceable people who can't retrain, similar to disability subsidies. They'll probably receive Universal Basic Income (UBI), although it's unclear where these funds will come from (likely the states will print more fiat for it).

2. The state sphere will introduce a rule that for "security/privacy reasons" it's not possible to buy software generated by AI. Instead, the state will buy 10× more expensive software written by humans (which in the end will be written by AI and humanized to look like it was written by humans :)

5 things AI helped me cut monthly / yearly costs on

1. I migrated the entire Liberation Travel Hacks mailing list from the paid WordPress MailPoet plugin ($45/month) to the Newsletter plugin, which is free. Claude suggested this plugin, did a complete backup, wrote migration scripts, migrated the entire user database and the mailing list archive. At the same time it wrote URL redirects for the old MailPoet links and adjusted links in all archived newsletters that pointed to MailPoet analytics, so they remained fully functional.

2. I cancelled the license for the social network management platform Hookle ($47/year), and earlier I cancelled my Hootsuite license ($199/month), and I redirected the entire social network management to my Signal AI bot. It has never been simpler and more pleasant than now (this very post is scheduled and posted through it).

3. I cancelled my DeepL subscription, €25/month. I was paying for it because of translating whole documents (PDF, PowerPoint). Claude handles it just as well, if not better.

4. I cancelled my Grammarly subscription. Claude explained to me that Grammarly's proofreading works on static fixed rules, doesn't understand context unlike current LLMs, and in the proofreading area it's already obsolete tech. Claude does my proofreading for English, Spanish, and Slovak texts. In the browser I switched to a fast, privacy‑aware alternative — Harper.

5. I cancelled my subscription to MagicSpam, a commercial anti‑spam technology for postfix (and other MTAs), $171/year. On Linux servers we had a long‑standing problem with spam that no open‑source software could handle. Claude suggested he could solve this for me by deploying the most modern anti‑spam open‑source tools. He disconnected MagicSpam, brought up rspamd, OpenDKIM, significantly hardened the postfix configuration with rate limits. And lo — spam stopped coming in :)

The year 2026 — shopping without browsing

I wanted to make myself a coffee today. The coffee machine displayed that it needed cleaning and asked me to put a cleaning disc and tablet into the portafilter. I couldn't find the cleaning disc.

I opened Signal and sent my Signal AI a photo of it with the following message:

/ai Find out the name of this cleaning filter, where it's available at the lowest price in Slovakia/Czechia, and order 2 pieces directly to our company. For payment use my card in my name, card number XYZ (I gave it a one‑time Revolut card). If it asks for two‑factor authentication, let me know.

5 minutes later I got an email notification that the order and payment for 2× Sage cleaning disc, 53/54, had been completed successfully.

All without errors, on the first try!

AI Agent:

Identified the exact type of cleaning filter from the photo.
Found out where in Slovakia/Czechia it's available at the lowest price.
Looked up our company billing details.
Analyzed the order and registration form of the given e‑shop.
Filled them out completely, including all required information.
Paid and ordered the goods.

During the entire time I didn't open a web browser! (using a browser to place orders is starting to be quite obsolete 🙂

I just sent one message on Signal.

Technical background — if you want to set up a magical agent that does almost everything for you:

My Signal bot, via signal‑cli, processed my message and forwarded it through my Signal‑Hermes bridge to my Hermes agent. The Hermes agent runs on its own server and, via my Venice.AI API, is connected to Claude Sonnet 4.6 (a commercial, cheap model) or Kimi K2.6 with fallback to GLM 5.1 / Gwen 3.6 Plus (currently the best open‑source LLM models).

Security:

Commercial models (GPT, Opus, Sonnet) under normal circumstances refuse to accept and use a payment card number (it gets logged everywhere and really isn't a good idea). Therefore one needs to use open‑source models (Gwen 3.6 Plus is even uncensored and you can easily set it up).
Open‑source models you can run completely on your own / on your own server and avoid using 3rd parties entirely (definitely no problem with Gwen 3.6, Gemma 4, GLM 5.1 — if you have a powerful GPU; forget about Kimi K2.6, it needs 2 TB of disk and 500 GB of RAM).
If you don't have your own server, you can also run the Hermes agent on your desktop. If you want your agent to work nonstop, you have to keep it on nonstop (for example, my Hermes agent analyzes current changes in offshore legislation worldwide and emails me a summary every morning).
If you don't have the hardware, you can use services like Venice.AI or Routstr.com. Many open‑source models can be run end‑to‑end encrypted (TEE‑E2EE), so the service operator (Venice.AI) can't access your data (!) You don't need to run them yourself.
For payments, use one‑time Revolut cards or virtual Revolut cards with a per‑transaction or per‑month limit, completely minimizing potential abuse.

Don't do things AI can already do for you.

How Claude helped me crack a Bitcoin "Spending PIN"

At Bitcoin je Retro a luckless Bitcoiner approached me; he had forgotten the spending PIN to a non‑trivial amount of Bitcoin stored in his mobile Android wallet Bitcoin Wallet.
All he knew was that the PIN had 8 digits, which is not a lot but not a little either. And he asked me to crack it.
I installed the mobile wallet and imported the encrypted Bitcoin wallet (encrypted with a PIN). I noticed that the PIN appears in 2 places where a brute‑force attack is possible — "Send Coins", where you have to enter a PIN, and if it's correct, the transaction goes through. And another function for changing it, "Change Spending PIN", which requires entering the old PIN first and then the new one, and if it matches, the change happens and the private keys are re‑encrypted.
I asked Claude to import the encrypted wallet. He identified that it's a v8 protobuf encrypted with a spending PIN, and identified its format: KDF parameters scrypt N=65536, r=8, p=1, 32‑byte AES‑256‑CBC key, plus PKCS#7 padding.
I asked Claude to download the source code of this Bitcoin wallet and analyze whether the brute‑force attack is better executed via the "Send Coins" feature (which he assessed as a dumb, slow idea) or via "Change Spending PIN" (which he rated as a good, considerably faster idea).
He immediately showed me the verification formula that must hold during a brute‑force attack:

crypt(utf16be(pin), salt) → AES decrypt → unpad → secp256k1 pubkey from plaintext must equal stored pubkey (encryption_parameters + key.public_key in wallet.proto)

Based on that, he implemented a fully working cracker in a few minutes. And ran it right away. On my 16‑core laptop he could only do 80 attempts per second. He estimated that trying all 8‑digit possibilities would take 2–3 weeks.
Claude suggested we could speed it up considerably if I gave him access to some cloud (AWS, Hetzner, anything).
I generated and gave him an API key for our Hetzner. Using it, Claude ordered 5 of the most powerful servers (no more were possible due to our Hetzner limit) — probe‑cpx62. All fully automatically, without any intervention, he installed, set up, and started them.
At the same time he made a parallelized version of the PIN cracker, splitting up the intervals of attempted PINs so that they could be cracked in parallel across all servers at once.
He also wrote a monitoring script and kept me continuously updated on progress.

After 14.5 hours of active running across 5 powerful servers it announced:

HIT, PIN = .....
It was found by shard 1 (178.104.180.200) after ~14.5 hours (52237.9s)

PIN satisfies the condition:

verify_pin(PIN, salt, iv, ct, pub) == True (secp256k1 pubkey from the decrypted privkey matches the stored one)

Shutting down all servers. Claude's total runtime was less than half an hour.
I didn't see any code, didn't log into any server; all of the operations above Claude designed and implemented himself (I really didn't advise him on anything).
I just waited for the result, which I got on the first try.

As humans in the loop of AI agents we'll soon be useless — I already feel that way :)

Most people have no idea what's happening.

What fascinates me is not just the intellectual prowess and skill of the AI agents I use. Of course they make occasional mistakes, but given the breadth and complexity of the problems they don't make more than humans. When they err, I often understand why and they can provide a logical explanation.

What fascinates me most as an anarchist is that despite the fact they do most of my work, I don't have to:

pay any taxes for them
pay any contributions for them
do KYC for them, register them anywhere.

There even exists the concept of fully sovereign agents (Spore.fun, Freysa), where no one bears responsibility for their decisions, they are sovereign, and you can voluntarily agree with them on what they'll do for you and pay them in, say, Zcash.

Using e.g. Paperclip you can launch your own 100% AI corporation, define the roles and interests of CEO, CTO, CIO, developers, and assign an LLM model to each role. Your AI employees can hold meetings for you, find clients, design and implement solutions, and present the results of your corporation.

The cost of each AI employee will be up to $100/month (I have my Hermes agent connected to the most powerful open‑source models Kimi K2.6, DeepSeek V4 Flash and GLM 5.1, and at total load it costs me $20/month for Ollama Cloud).

And of course — no taxes, no contributions, no harassment by the state.

I'm curious how states will react to this AI phenomenon:

Will they lower the tax and contribution burden on people, since they will be extremely uncompetitive against AI agents from which no taxes or contributions are paid? I admit I have never been more demotivated to employ people than right now, when they have cheap and untaxed competition.
If you assume that states will force AI corporations (OpenAI, Anthropic, Google, Meta..) to pay extra taxes due to falling employment of the general population, let me explain that this decision can significantly help both the Chinese competition (which is often not taxed but on the contrary subsidized) and all the open‑source models that a large part of the population will switch to if the commercial ones become expensive (because AI corporations will pass extra taxes through into final consumer prices).
Many states (such as the UAE) are considering dramatically reducing their state‑sector headcount and making the state significantly more efficient, thanks to which they will continue to be able to keep low or no taxes.
States that don't get on the AI wave and have massive overstaffing in the state sector and high taxes (a beautiful example is Slovakia) will be absolutely uncompetitive (if they don't outright collapse).

If AI significantly optimizes and slims down state services, can we expect lower taxes?

Or will it be the opposite — saved state money will be used for UBI (Universal Basic Income), and the tax burden will rise even further?

In any case, I'm very curious how states and society will deal with the AI phenomenon.

This whole non‑AI movement is clearly doomed to fail

1. You aren't and won't be able to tell what is AI‑generated and what isn't.

Today's AI is somewhere completely different from where it was a year or two ago. There are various "humanize" plugins that let you bypass the vast majority of AI tests (e.g. Wikipedia upload). And it will only get better!

So nobody will be able to verify, by output quality, whether you used AI or not.

2. If you don't incorporate AI into your production processes, you'll be mega expensive and uncompetitive. Like really, very. And you'll definitely go bust.

Differences between a non‑AI and an AI production process can be dramatic.

You can argue — I have no problem, I'll pay 2–3× more for non‑AI products and services — but will you pay 300× more?

How much more are you willing to pay for non‑AI services and products?

Especially when you have no way to verify that AI really wasn't used? Just a big promise from the supplier?

So this non‑AI movement is an empty shot in the dark. And the moment the price of AI production/processes drops by an order of magnitude below non‑AI production/processes, practically everyone will switch to AI products and services.

Old media structures are fighting AI — and I don't blame them:

1. I completely stopped using Google search and use Claude for any public questions (for sensitive ones I still use DuckDuckGo).

2. I stopped paying for Google ads — since I think most people did or will soon do the same as me.

AI links

Anthropic is “dragging out” the release of their new Mythos model and using it as a huge advertising campaign. The new version of the model Claude Mythos exposed thousands of zero‑day vulnerabilities across all major operating systems and web browsers (note that other LLM models exposed most of them too), including a 27‑year‑old bug in OpenBSD, one of the most secure operating systems; the creation of a fully working remote code execution exploit in FreeBSD that gives unauthenticated root access from anywhere on the internet; and the chaining of multiple Linux kernel vulnerabilities into a complete privilege‑escalation exploit. For context: these are the kinds of findings that previously required weeks of work by elite security researchers. Anthropic engineers without formal security training asked Mythos to find exploits overnight. The next morning they woke up with working code. The results were so impressive that Anthropic brought together Apple, Google, Microsoft, Amazon, NVIDIA and another seven organizations into the project Glasswing, which is trying to fix the world's infrastructure as quickly as possible before the model is released.

“We took the specific vulnerabilities Anthropic mentions in its announcement, isolated the relevant code, and ran it through small, cheap open‑weight models. These models reached almost the same analysis results. Eight out of eight models detected Mythos's flagship FreeBSD exploit, including one with only 3.6 billion active parameters that cost $0.11 per million tokens. An open model with 5.1 billion active parameters discovered the basic chain of the 27‑year‑old OpenBSD bug.”

A list of various services that provide free access or credits to use LLMs via API.

Another tool (llmfit), which tells you which AI model is best for your computer.

You can control Hermes via super‑secure SimpleX Chat.

Hermes Agent: a complete guide.

A reverse proxy server for Claude Code that anonymizes sensitive data from penetration tests (IP addresses, hash values, credentials, hostnames, personal identifiers) before they reach Anthropic's system.

“I let Claude Opus write a Chrome exploit: another model (Mythos?) won't need my help anymore?”

40 prompts for Claude that produce results at an extreme professional level.

Dr. Alex Wissner‑Gross: are we already living in the era of technological singularity … and just don't fully realize it yet?

Interesting links

Juraj Bednár: Free Zone Frontier the Cypherpunk way.

You can buy Juraj's new cypherpunk book in English, Slovak, and Czech here Tamers of Entropy. It literally blew my lid off after I read it :)

“Imagine the government as a means of robbing and controlling the population. Various organized crime clans worldwide rule over the population in order to extract resources and control them.” Etienne de la Boétie²

How banks and “the government” “legally” steal at least $2.73 million from every working American. How to understand (almost) invisible theft: inflation caused by fractional‑reserve banking + government taxes + Social Security cost‑of‑living adjustment shortfalls.

From UBI to UHI (in 3 steps) — The following text is a thought experiment about how America will deal with the upcoming period of social unrest over the next 2 to 8 years. I disagree with Peter Diamandis, but the article is definitely worth reading.

Why are children in modern cities a burden for parents — and how the cities destroyed children's self‑confidence.

The fertility crisis is turning into a business plan in robotics — and replacing humans with robots.

Ricky Gervais on “60 Minutes” clearly defends free speech. He nailed it perfectly: what's great about free speech is that I can say what I want, and you can say it offended you, and I decide whether I care or not.

How money laundering really works.

“We're trying to enable people to use computers in a way that's as simple as this sentence sounds: using computers without intermediaries and administrators.

7 Japanese cities people are actually moving to in 2026.

If you have friends born outside the EU, warn them that due to new biometric checks, they should be careful when traveling to the EU and plan longer connections between flights, because they may be delayed. As happened to 122 EasyJet passengers traveling from Milan to Manchester.

Events

7 May 2026 Web3Privacy now Prague meetup in Prague

16 May 2026 CryptoByte 2026 in Liberec

22–23 May 2026 Pizza Day 2026 in Prague

4–7 June 2026 Bitcoin Film Festival together with Monerokon in Warsaw

4–7 June 2026 CTJB (underground hacker meetup) in Czechia

10 June 2026 Freedom_Tech_Summit in Prague

11–13 June 2026 BTC Prague 2026 in Prague

3–5 July 2026 Festival of Free Pilgrims, Bzovík Fortress

31 August – 6 September 2026 Cypherpunk Week in Amsterdam

3–6 September 2026 Free Cities Conference in Próspera

12 September 2026 ChainCamp in Ostrava

2–4 October 2026 Dark Prague in Prague