How to use AI positively (and carefully)

If you are a creative, don’t run away just yet!

Actual real art is being made using AI by people like Refik Anadol:

Table of contents

Introduction

Other than this introduction and the section below on the coming US Brain Drain, this article was prepared entirely by AI (Gemini Deep Research) without any edits by me. For an example of how these Reports look when they are created and their depth of research see:

I used a prompt that was not country specific, but you can – of course – ask the tools to focus on particular countries, sectors, entrepreneurs and businesses.

You will note the response of the AI tool below is still skewed towards the USA being considered a safer less authoritarian jurisdiction. That is not a safe assumption, but it is useful as it highlights how Large Language Model AIs are inherently backward looking (due to their training data and the larger corpus of information available further back in time).

That said, we humans are all guilty of this skew too, which can make it difficult to adjust in rapidly changing times. After all, we are walking backwards into time, all we see is history in front of us.

As I have not edited the article (other than by adding headings) I have also not checked the sources, so please check all sources carefully if relying on this article for anything. You can also check any AI produced factual outputs against a different AI tool (with a different engine / foundation model) to see if they spot any errors.

If I was doing a more detailed research article, I would add more details on US specific risks and also do a deep dive on topics like Starlink, X, Palantir (see example Perplexity Deep Research report here), spyware and applicable anti terrorism laws that could be misused to access your data and activities. I would then collate human and AI generated sources to prepare a human generated article.

You can also ask AI programs for help finding project management tools and in managing tasks, step plans and safe coordination in groups.

I think we will soon see an exodus of US tech and AI firms to Europe.

Will Europe Benefit From A US Brain Drain?

Whilst we might wish it were not reality, the Project 2025 crisis and a deeply extremist right wing US administration is a real opportunity for the Europe (including the UK) to benefit from political and social destabilisation in America at the moment.

I think we will soon see an exodus of US tech and AI firms to Europe. You would expect AI developers will not want to move to America or stay there when even Fulbright scholars are being snatched from the streets in broad daylight by ICE agents¹ and taken to undisclosed locations without access to legal support (or even being charged with a crime).

Other people are being deported to a maximum security hell-hole prison in El Salvador for having tattoos² and again with no due process (including American citizens).³

“It was exile to another world, a place so cold and far from home they may as well have been sent into space, nameless and forgotten. Holding my camera, it was as if I watched them become ghosts.”⁴
(Philip Holsinger)

Many American technology and AI businesses with ethical standards are going to need to plan their exit from America very soon – if they want to protect their customers and brand. None of them can honestly tell you they will be able to protect you and your data from a deeply authoritarian increasingly fascist US Government.

There are also economic reasons why US based tech companies will find being aligned with the Trump administration against Europe painful.⁵

Europe, including the UK, have the strongest data protection, AI and social media regulations in the world. This is a competitive advantage in an increasingly authoritarian world. The genesis of this strong European data protection law is the Holocaust and widespread surveillance by the Nazis, Stasi and others.⁶

We should expect an American brain drain to occur across a wide range of scientific fields – a real opportunity for Europe ⁷.

Data sharing for intelligence agencies and businesses between the EU/UK and the USA-Israel need to be significantly modified or halted, further segregating these traditionally close economic blocs.The USA-Israel can not meet European data protection adequacy standards.⁸⁹ Extradition to the USA will also now be much harder from Europe given the human rights compatibility test.¹⁰ This also provides a competitive advantage for technology developers worried about the long arm of US law.

We are beginning to see socially conscious American technology firms take a stand against the Trump administration’s attack on immigrants and vulnerable people, these firms will likely need to leave American soon.¹¹

Project 2025

Anyway, I hope this is helpful to give you ideas of how AI can be used in your research projects, for citizen journalism or for citizen support groups you may be involved in.

You might also want to mashup different AI tools for example:

Ask for Deep Research on the same or similar topics from two different AI agents; then
Save the results into your AI notebook;
Use that notebook tool to query those results and any other articles/documents you add manually (notebooklm Pro allows for 300 sources)

As with nearly all technologies (language, culture, the steam engine, atomic power, the internet, computers) these tools are not inherently evil. What matters is how we try to use technology (including the words we use) if they are to be used as a force for positive social change.

Remember, we can’t beat the technofascists (with their much greater access to resources) if we don’t use all the technology tools available to us. We will be overwhelmed with data (as they flood the information zone –usually with shit) and we could end up feeling helpless or disheartened with trying to make sense of the world.

Please also be very careful and only use AI tools where you know who owns them, where the servers are located that deliver the services and what access your government or another government may have to the data. Obviously do not use Xai or Grok!

If you are at a higher risk of surveillance or arrest (sadly that covers many people in many countries) then you should read this summary of some of the key security issues including the risk of using cloud services and American based AI tools.

Your use of AI could be discovered or made available the same way as your internet search, so great caution and careful research on the best AI tools to use is needed!

The misuse of draconian and wide ranging anti-terrorism and security laws or policies to access your devices and information must not be underestimated.

AI as a Force for Good: Resisting Authoritarianism and Countering Technofascism

Artificial intelligence stands as a transformative technology, wielding the power to reshape societies in profound ways. This power, however, is double-edged.

While offering unprecedented opportunities for progress and innovation, AI is also increasingly being employed by authoritarian regimes to tighten their grip on power.

From sophisticated surveillance systems to automated censorship and the spread of insidious disinformation, AI is becoming a cornerstone of modern digital authoritarianism.

This trend has fueled concerns about the rise of “technofascism,” a system where technology is used to centralize control, suppress dissent, and erode fundamental civil liberties. Yet, within this landscape of concern lies a powerful counter-narrative: the potential for AI to serve as a force for good, empowering individuals and groups to resist oppression, access critical information, and organize for change in environments where freedom of expression is curtailed.

This article explores both sides of this complex issue, delving into how authoritarian regimes misuse AI and, more importantly, outlining the practical ways in which individuals can leverage privacy technologies and AI tools to safely research issues, find hidden data, and synthesize information in their fight for a more open and just world.

Misuse of AI by Authoritarian Regimes

Authoritarian governments across the globe are rapidly integrating AI into their strategies for maintaining control ¹. This integration manifests in several key areas, each designed to enhance the state’s ability to monitor, silence, and manipulate its citizens. One of the most prominent applications is in the realm of surveillance.

AI-powered facial recognition technology, particularly prevalent in countries like China, enables the real-time tracking of individuals’ movements with remarkable efficiency ¹.

With extensive camera networks augmented by AI, governments can identify and monitor individuals as they move about, creating a pervasive sense of being watched ¹.

This constant digital scrutiny fosters an environment of fear and self-censorship, where citizens become hesitant to voice dissenting opinions or engage in activities perceived as contrary to the state’s interests. The normalization of such surveillance, starting from a young age, could even lead to future generations internalizing conformity as the norm ¹.

Beyond mere tracking, authorities in some regimes aim to utilize AI to predict individuals’ movements and behaviors, framing this as a measure to prevent crimes ¹. However, this predictive policing carries the significant risk of devolving into profiling and the silencing or even conviction of individuals for “uncommitted criminal activities,” raising profound concerns about the erosion of privacy and fundamental freedoms ¹.

The efficiency of AI in mass surveillance also presents a concerning development, potentially decoupling the need for large, visible police forces from the act of maintaining control ².

This makes surveillance a more sustainable and potentially less overtly violent tool for authoritarian regimes, allowing them to exert control and deter unrest before it even becomes visible ³.

Furthermore, the global spread of AI surveillance technology is being driven significantly by countries like China, which, through initiatives such as the Digital Silk Road, are becoming major exporters of these tools to developing and authoritarian states ⁴. This creates a network of nations employing similar technologies and tactics for digital authoritarianism, posing a widespread threat to democratic aspirations worldwide ¹.

AI for Censorship

Censorship is another critical area where authoritarian regimes are leveraging the power of AI.

AI algorithms are being employed to automate and refine the censorship of online content, including text, images, and videos, with increasing speed, scale, and subtlety ⁴. This makes it considerably more challenging for dissenting voices to reach an audience and for citizens to access information that the state deems undesirable ⁶.

Technically advanced authoritarian governments are also actively working to ensure that even advanced AI chatbot applications adhere to their censorship systems ⁶.

This proactive approach aims to control the narratives generated by these increasingly popular AI tools, preventing the dissemination of information that contradicts the state’s agenda ⁶.

Democratic openness, paradoxically, could even aid authoritarian regimes in enhancing their censorship capabilities, as AI trained on data from free societies might learn to identify subtle forms of dissent that are also present in more controlled environments ¹⁰.

A particularly alarming trend is the use of AI models developed in authoritarian states to actively censor historical events, deny human rights abuses, and propagate state-sponsored propaganda ⁸. This systematic distortion of information represents a significant threat to the integrity of the global information space and the ability of individuals to access factual accounts of sensitive topics ⁸.

Deepfakes, Misinformation & Disinformation

Beyond surveillance and censorship, AI is playing a crucial role in the spread of disinformation and the manipulation of public opinion for social control.

Generative AI technologies are supercharging disinformation campaigns through the rapid and cost-effective creation of highly realistic fake images, videos, and audio, commonly known as deepfakes ⁴.

The increasing sophistication and accessibility of these tools lower the barrier for both state and non-state actors to engage in widespread information manipulation, making it difficult for individuals to discern authentic content from fabrications ⁴.

AI-powered chatbots are also being harnessed to disseminate falsehoods across the internet at unprecedented speeds, often utilizing machine translation to overcome language barriers and spread propaganda and extremist agendas, reinforcing existing biases within echo chambers ⁴.

These applications can be combined to automate the entire process of synthetic content production, distribution, and amplification, posing significant risks to our understanding of the world and our trust in information sources ⁴.

Furthermore, some authoritarian regimes are exploring the use of AI to predict individuals’ movements and behaviors, not just for crime prevention, but also as a means of social control, potentially incentivizing regime-compliant behavior and penalizing any form of dissent ¹.

Technofascism

The convergence of authoritarian governance and advanced technology, particularly AI, has given rise to the concept of “technofascism” ¹³.

This system represents a form of rule where technological expertise is deeply intertwined with political control, allowing for more subtle yet pervasive forms of societal management compared to traditional fascism ¹⁵.

Several core attributes characterize technofascism, and AI plays a complicit role in enabling them. Mass surveillance, facilitated by AI-powered cameras, facial recognition, voice analysis, and extensive data mining, allows for the constant monitoring of citizens, often justified under the pretense of ensuring public safety or national security ¹⁵.

This creates an environment where privacy is severely curtailed, and the potential for state overreach is significantly amplified ¹⁵.

Behavioral control is achieved through the analysis of vast datasets by advanced algorithms, which can predict, influence, or manipulate individual and collective behavior, subtly encouraging “acceptable” actions and discouraging dissent ¹⁵.

AI also powers sophisticated censorship and information control mechanisms, filtering and suppressing online content to shape public discourse and limit the expression of dissenting opinions ¹⁵.

The close alliances between governments and private technology companies, a hallmark of technofascism, create monopolistic control over crucial sectors like communication and data storage, often operating without adequate public accountability ¹⁵.

Moreover, AI systems, frequently built upon biased data, can enforce algorithmic discrimination and inequality in critical areas such as law enforcement and access to public services, perpetuating existing societal biases under the guise of objectivity ¹⁵.

Ultimately, technofascism leads to an erosion of privacy and autonomy, with constant surveillance and algorithmic decision-making undermining individual freedoms and the ability to act without being monitored and judged ¹⁵. Examples of these tendencies can be seen in China’s extensive AI surveillance infrastructure and its social credit system ¹⁵.

Concerns have also been raised regarding initiatives like the Stargate Project, involving collaborations between powerful tech figures and political leaders, as potential indicators of a drift towards technofascism in other parts of the world ¹⁵. These examples underscore the real-world implications of the increasing fusion of authoritarianism and advanced technology.

AI as a tool for resistance

Despite the concerning ways in which AI is being misused, the technology also presents a powerful arsenal for individuals and groups striving to resist authoritarianism.

AI tools and techniques can empower those living under oppressive regimes in several crucial ways, particularly in accessing information, organizing, and conducting research.

One of the most significant areas is information access and synthesis. In the evolving landscape of information warfare, advanced AI search systems built on open, reasoning-focused models like DeepSeek R1 are beginning to serve as powerful instruments for contextual synthesis. By integrating live web access with sophisticated reasoning—capable of parsing, correlating, and validating data from diverse global sources—these systems lessen dependence on centralised narratives and streamline the human effort required to interpret complex data and events. When coupled with privacy-preserving infrastructure they form a complementary ecosystem that enhances both individual autonomy and the resilience of collective information ecosystems.”²¹.

By hosting open-source language models on servers outside the jurisdiction of authoritarian governments, these platforms can provide uncensored access to information, even on topics that are typically restricted ²¹. This offers a vital alternative to state-controlled narratives, allowing individuals to form their own understanding of events based on a broader range of sources.

Complementing these AI-driven solutions are traditional privacy-enhancing technologies like Virtual Private Networks (VPNs) and Onion Routing (Tor) ²⁶. These tools remain essential for masking IP addresses, encrypting internet traffic, and providing a layer of anonymity that is crucial for secure online research and communication in environments where freedom of expression is limited ²⁶.

Additionally, AI content bypass tools can help users overcome geographic and political restrictions to access content that might be blocked in their region, further promoting the free flow of information ²⁶.

Encrypted communications

In the realm of organization and communication, AI is also proving to be a valuable asset for resistance movements.

Encrypted messaging apps like Signal, which often incorporate AI for enhanced security, provide secure channels for communication, protecting activists and individuals from government surveillance through end-to-end encryption ⁴².

AI is also being used to develop tools that can help activists enhance their overall digital security, detect the presence of spyware on their devices, and even debunk disinformation campaigns launched by authoritarian regimes ⁴². Furthermore, technology is facilitating safer and more informed participation in protests.

Platforms like Safe_Protests, a web application, offer features such as live updates, information on protester rights, and the ability to mark dangerous locations on a map, enhancing the safety and organizational capabilities of those engaging in peaceful assembly ⁴⁸.

For individuals and groups dedicated to uncovering and documenting human rights abuses, AI offers powerful tools for research and analysis.

AI algorithms can analyze large datasets of information, including social media content, news reports, and government records, to identify patterns and anomalies that might indicate human rights violations, often with greater speed and efficiency than human analysts ¹⁹.

The field of Open Source Intelligence (OSINT), which involves gathering and analyzing publicly available information, is being significantly enhanced by AI. AI-powered OSINT tools enable researchers and journalists to safely access and document human rights violations using data like satellite imagery and social media content, even in areas where physical access is restricted or dangerous ⁵⁶.

Moreover, AI tools are proving invaluable in analyzing leaked government documents, helping to sift through vast amounts of data to uncover evidence of corruption, human rights abuses, or other illicit activities that authoritarian regimes attempt to keep hidden ⁶¹.

AI platforms to evade geo-restrictions & use of VPNs

For individuals seeking to resist authoritarianism through information gathering and dissemination, taking careful steps to ensure their safety and privacy is paramount.

Utilizing privacy technologies is a fundamental first step. Choosing a reputable VPN provider is crucial. Services like NordVPN, Surfshark, and ProtonVPN are known for their strong encryption protocols, strict no-logs policies, and servers located in jurisdictions that respect user privacy ²⁷.

When selecting a VPN, it is important to consider factors such as the encryption standard used (ideally AES-256), whether the provider has undergone independent audits of its no-logs policy, the number and location of available servers, and the legal jurisdiction in which the company operates.

Once a VPN is chosen, using it safely involves downloading and installing the application on all relevant devices, creating an account, and connecting to a server located in a country with robust internet freedom before engaging in any online activities that could attract unwanted attention ³⁸.

For an even greater level of privacy, individuals can explore Onion Routing (Tor) through the Tor Browser, which encrypts and routes traffic through multiple volunteer-operated servers, although this often comes with slower connection speeds ³².

While smart DNS and proxies can also be used to bypass geographical restrictions, they typically offer less comprehensive privacy and security compared to VPNs and Tor ⁶⁷.

Unlocking access to information that authoritarian regimes attempt to suppress can be facilitated by AI-powered search engines. Platforms like Perplexity AI, especially when utilizing the DeepSeek R1 model, can provide uncensored search results, even on sensitive topics ²¹.

Users can often select these models within the platform’s interface and formulate queries on topics that might be heavily filtered by traditional search engines. Employing semantic search techniques, which focus on the meaning and context of a query rather than just keywords, can also help uncover less obvious information ⁶⁹.

Natural Language Search and Summaries

Asking natural language questions and exploring the advanced features of AI-powered search engines can yield more comprehensive and relevant results.

For more in-depth investigative research, exploring AI-enhanced OSINT tools like Maltego, Spyse, and Aleph Search (if accessible) can help gather and analyze publicly available data to reveal hidden connections and insights ⁵⁶.

Once relevant information is found, AI can significantly enhance the efficiency of knowledge synthesis. For researchers dealing with large documents, AI summarization tools such as QuillBot, Notta AI, Jasper, Hypotenuse AI, TLDR This, and Paraphraser.io can quickly condense lengthy articles, reports, and research papers into more manageable summaries ⁷⁸.

These tools typically allow users to upload documents or paste text to generate concise summaries. When dealing with information from multiple sources, the concept of “synthesis AI” becomes valuable. Tools like HyperWrite AI’s Literature Synthesis Assistant and Otio are designed to combine insights and findings from various documents, creating a cohesive narrative that synthesizes the information ⁸⁹.

For those working with academic research papers, Large Language Models (LLMs) can be particularly useful for extracting key findings, identifying relationships between concepts, and generating summaries ⁹⁴.

Among the growing array of AI tools, Deep Research and notebook LLMs stand out for their potential to aid individuals resisting authoritarianism.

Deep Research

Deep Research, as seen in features offered by Gemini and ChatGPT, represents a significant advancement in AI’s ability to autonomously conduct multi-step research on complex topics ⁹³.

These tools can browse hundreds of websites, analyze the information, and synthesize it into comprehensive reports with clear citations, saving researchers countless hours ⁹³. This capability can be particularly valuable for in-depth investigations into government activities, allowing individuals to quickly gather and process information from a wide range of online sources ⁹⁸.

Deep Research can also be instrumental in uncovering niche or non-obvious information that might be missed through traditional search methods ⁹³.

Notebook LLMs, such as Google’s NotebookLM, offer a different yet equally powerful approach ¹⁰⁶. These tools function as AI-powered notebooks where users can upload various sources, including documents, web pages, audio, and videos, and then use AI to synthesize and analyze the information ¹¹⁰.

Notebook LLMs excel at summarizing uploaded documents, answering questions based on the content with inline citations, and even generating study guides or briefing documents ¹¹⁰.

For individuals in restrictive environments, the ability to upload and analyze local copies of documents or information accessed via VPN makes Notebook LLMs particularly useful for working with potentially censored materials ¹¹¹.

The “Audio Overview” feature offered by some notebook LLMs provides an innovative way to engage with research material by listening to AI-generated discussions about the uploaded sources, which can be beneficial for auditory learners or for multitasking ¹⁰⁷.

Conclusion

In conclusion, the relationship between AI and authoritarianism is complex and multifaceted. While authoritarian regimes are increasingly leveraging AI to enhance their control through sophisticated surveillance, censorship, and disinformation campaigns, the same technology holds immense potential as a force for good in the hands of those who seek to resist oppression.

By embracing privacy-enhancing technologies and strategically utilizing AI-powered search engines, data analysis tools, and knowledge synthesis platforms, individuals can gain access to critical information, organize more effectively, and document human rights abuses, even in the face of significant restrictions.

The development and accessibility of human-centered AI tools that prioritize privacy and security are crucial in empowering individuals in their fight for a more just and open world. As technology continues to evolve, it is imperative that individuals recognize their agency in harnessing these powerful tools responsibly and strategically to challenge authoritarianism and work towards a future where information is freely accessible and fundamental human rights are universally respected.