You are naked: Your smartphone is open – for anyone who can pay
We have practically nothing left that is truly private and hidden. Many people think that artificial intelligence has only recently become a major topic – since ChatGPT and similar programs became widely known. But governments and intelligence agencies have been using AI intensively for many years, in some cases for over 15 years. This gives countries like the USA, China, and Russia a huge advantage that most people don’t fully understand.
The massive data theft by China
In 2015, China stole the personal data of over 21 million US government employees in the so-called OPM hack. Those affected were primarily active and retired soldiers, intelligence officers, and civil servants, but also their families and acquaintances. Extremely sensitive data was stolen: social security numbers, fingerprints (of 5.6 million people), addresses, financial information, detailed histories from security clearances (SF-86 forms), information about family, friends, foreign travel, and even psychological information.
This data is invaluable for espionage, blackmail, and long-term intelligence operations. The hack was largely attributed to China – specifically the Ministry of State Security and the Jiangsu State Security Department. There were arrests (e.g., of a Chinese national for using malware) and official US statements blaming China.
TikTok collects massive amounts of data
It has hundreds of millions of users worldwide. The app collects an enormous amount of information: location data, contacts, photos, videos, and even what users type within the app. Researchers have shown that TikTok’s built-in browser reads keystrokes – including passwords, credit card numbers, and other sensitive text. Because TikTok belongs to a Chinese company, and Chinese law allows the state access to company data, there is a risk that this information could eventually end up in the hands of the Chinese government.
Every device can be remotely accessed
Once you become a target – whether for political, economic, or criminal reasons – there is hardly any protection. Spyware exists that allows you to completely access and read the data of smartphones, laptops, tablets, or even smart TVs remotely, without needing to be nearby. Such attacks often occur without a single click. Spyware like Pegasus from the Israeli company NSO Group or Predator from Intellexa enables zero-click exploits: A single invisible message is all it takes for the attacker to have full access to the camera, microphone, messages, photos, and files.
Some of these tools cost between a few thousand and tens of thousands of dollars for a single target device – depending on the quality and the provider. There are commercial providers that sell such services, and similar options are available on the black market. Often, all you need is a phone number or a close contact who has the target’s app. The cost? For Pegasus, they start at $500,000 setup plus per target device, but there are reports of cheaper options on the black market, often in the five- to six-figure range, which is plausible for simpler tools.
The US regulatory authority FTC penalized data brokers such as X-Mode or InMarket for selling sensitive location data.
Apps share data simply through proximity
When two phones are close to each other, they automatically exchange data via certain apps – even without the user noticing. Apps from meta-platforms (Facebook, Instagram) in particular use location services, Bluetooth, Wi-Fi, and other signals to detect who is nearby. This is called geofencing or proximity sharing. As a result, it’s sometimes possible to find out who is near whom, even without a phone number, and thus link their data.
AI was new to the public – but not to intelligence agencies
While most people only really came into contact with generative AI in 2023/2024, intelligence agencies like the CIA have been using AI for many years – back then not for fancy videos or deepfakes, but for analyzing huge amounts of data, pattern recognition, and predictions. This head start of 15 years or more means that the actual capabilities of countries like China or the USA are probably far greater than what is publicly known.
As early as 1999, the CIA founded its own technology investment arm, In-Q-Tel, with the goal of rapidly adapting new technologies from the civilian market for intelligence purposes. By 2006, In-Q-Tel had already funded over 90 companies and introduced more than 130 technologies into the US intelligence community.
A well-known example is Palantir. The company was co-financed by In-Q-Tel starting in 2004/2005. Palantir developed software that could link, analyse, and search for patterns in huge amounts of data from different sources – precisely the capabilities that are now summarized under “AI-powered analytics”.
Between 2005 and 2008, the CIA was one of the key early adopters of this platform. The software was used to visualize relationships in complex datasets, analyse networks, and more quickly identify indications of security-relevant activities . The fact that AI and machine learning have now become an integral part of its strategy is also demonstrated by the growing focus on AI in In-Q-Tel’s current portfolio. This development, therefore, did not occur abruptly, but rather over more than two decades.
The widespread impression that government use of AI only began with modern language models is therefore incorrect. Data analysis, pattern recognition, and algorithmic evaluation have been part of the operational business of the US intelligence apparatus since at least the mid-2000s.
How can you even protect yourself anymore?
In today’s world, almost nothing is truly private once you’re targeted. Nevertheless, there are actions you can do: delete unnecessary apps, severely restrict permissions, only enable location services when needed, use a VPN, and regularly check your devices. Authorities like the FTC in the US are now cracking down on data brokers who sell sensitive information. But the best protection remains: leaving as few digital traces as possible.
yogaesoteric
March 11, 2026