View this post on the web at [link removed]

By Trygve Olson
This is the second part of the series, AI vs. Autocracy: Seven Lessons in the New Battle for Democracy.
In every authoritarian system I’ve worked in — from Belarus to Azerbaijan to Zimbabwe — opposition voices weren’t just suppressed.
They were tracked. Profiled. Punished.
And now, with large language models (LLMs) and machine learning tools becoming more powerful by the day, that process is going from manual to automatic — and from subtle to systemic.
This isn’t a future threat.
It’s happening now.
Authoritarians no longer need to monitor everyone.
They just need everyone to believe they’re being monitored.
AI does that faster than any human intelligence service has ever been able to.
What the Autocrats Will Do
AI — especially LLM-powered platforms and pattern-recognition systems — allows regimes to automate repression:
Facial recognition systems at protests, political events, or polling stations
Sentiment analysis of social media to flag “unpatriotic” speech
Predictive policing tools that identify and preemptively disrupt activist networks
Automated employment blacklists based on online behavior
Loyalty scoring systems that punish quiet dissent with visa denials, blocked loans, or frozen benefits
AI-enhanced data aggregation that tracks not just what you say — but who you talk to, when, and how often
It’s not about what you’re doing.
It’s about who you are — and how easily you can be profiled.
In The Battle for Democracy, I wrote:
“Authoritarian systems don’t need to control everyone. They only need to convince dissidents they are being watched. Fear is a far more efficient enforcer than force.”
AI doesn’t just scale surveillance.
It scales self-censorship.
What Democracy Must Do
We can’t stop this by legislation alone, though regulations will matter.
We must use AI to defend the rights it’s being weaponized to destroy.
That means:
Building secure, AI-assisted communication tools for organizers, journalists, and communities under pressure
Training civic actors — from city councilors to pastors to poll workers — to recognize the signs of AI-driven targeting and disinformation
Developing early-warning systems that detect digital targeting of dissent before it becomes state action
Creating off-ramps: ways for people who are being watched, threatened, or flagged to access help — without needing technical expertise
And above all:
We need to defend not just speech, but the act of speaking.
If autocrats use AI to isolate voices, democrats must use it to protect and amplify them.
Three Things You Can Do Today
1. Support digital civil society tools.
Platforms like Signal, Tella, and CivitAI are building tools to keep activists and everyday citizens safer. Donate. Share. Use them.
2. Push for transparency in local surveillance systems.
Ask what data your city, school district, or state collects — and how it’s used. AI-based systems often operate in secret. Shine a light.
3. Normalize secure communication.
You don’t have to be a dissident to use encrypted apps. The more normal it becomes to communicate securely, the harder it is to target only the vulnerable.
Bottom Line
Autocrats don’t fear AI.
They see it as the ultimate loyalty test.
That’s why this moment demands more than concern.
It demands a firewall built not just with code, but with conviction.
Because if AI can be used to silence, it can also be used to shield.
Let’s make sure that’s what we build — before the window closes.
Trygve Olson is a strategist, pro-democracy fighter and a founding Lincoln Project advisor. He writes the Searching for Hope [ [link removed] ] Substack. Read the original column here [ [link removed] ].

Unsubscribe [link removed]?