They’ll Use AI to Engineer Division. We Must Use It to Strengthen Solidarity.
Authoritarians will use AI to simulate solidarity — and sabotage it.
By Trygve Olson
This is the third in a seven-part series: AI vs. Autocracy: Seven Lessons in the New Battle for Democracy.
In fear-based systems, division is the fuel.
Autocrats don’t need to be loved.
They need people to distrust each other enough to give up on cooperation.
Now, with AI — especially large language models (LLMs) trained to mimic human tone, logic, and cadence — they can manufacture division at scale, in real time, and with increasing surgical precision.
This isn’t ideological.
It’s psychological.
And it’s already happening.
What the Autocrats Will Do
AI tools allow regimes and bad actors to:
Generate disinformation campaigns aimed at turning allies against one another
Inject content that appears to come from trusted voices, but is fake
Flood community threads with identity-based attacks to provoke fracture
Deepfake or impersonate opposition figures to break coalitions from within
And we’re not speculating.
Last month, it came out that a fake Signal account impersonating Secretary of State and Acting National Security Adviser Marco Rubio contacted at least two members of Congress and a foreign leader.
The messages were convincing. They were written in Rubio’s voice.
The objective? Not propaganda — division.
To undermine diplomacy. To breed confusion. To test how easily trust between allies could be fractured with AI.
That wasn’t a prank. That was a trial run.
And if the Secretary of State can be spoofed that easily, how do you know the person in your coalition group chat is really who they say they are?
What Democracy Must Do
We don’t fix this by ignoring it.
We fight division with intentional solidarity — built with trust, tools, and vigilance.
That means:
Using AI to detect and map coordinated disinformation campaigns
Building alert systems to warn movement leaders when impersonation or identity-wedge tactics are being deployed
Training grassroots and national coalitions to expect this kind of infiltration — and remain anchored anyway
The next wedge won’t come from the outside.
It will realistically appear as if it originated from within your team.
They’ll Use AI to Target Dissent. We Must Use It to Protect Expression.
Authoritarians no longer need to monitor everyone. They just need everyone to believe they’re being monitored.
Three Things You Can Do Today
1. Watch for patterns, not just posts.
If a message triggers infighting, ask: Who benefits from this fracture? Then check the source.
2. Build real relationships offline.
Digital trust can be gamed. Personal trust takes longer and holds stronger.
3. Establish communication norms before crisis hits.
Agree in advance: How do we confirm messages? Who speaks for the group? What’s the backup channel?
Bottom Line
Authoritarians will use AI to simulate solidarity and sabotage it.
We must utilize both AI tools and human strategy to reinforce what is real.
Solidarity isn’t soft.
It’s a security system.
And in 2025, division isn’t just a risk — it’s a weapon.
Trygve Olson is a strategist, pro-democracy fighter and a founding Lincoln Project advisor. He writes the Searching for Hope Substack. Read the original column here.
 | A guest post by
|
Sadly, dim democrats are so behind the the AI tech boom, and tech bros are all in for the kleptocrat trump that it's like starting a 100 meter dash, 10 meters behind your competitor. Dems were once on the forefront of technology, and now they are still writing strongly worded letters. Maybe Lincoln Square and its tech savvy group could invite old guard, and I mean old, to a How to Use a Computer seminar.
Newsom seems to have gotten the message, and is fighting to musk grok crock AI with the same intensity, although it is more mockery based, and not 1984 based, precision. It is one of the best attempts to slap down the orange bloat and his maga morons with their own drivel. Keep up the good fight, Lincoln! Your efforts are important and appreciated.
Watch for patterns. Good advice. Patterns that go sideways are telling. Ask who benefits. Means we do need to be persistent watcher and listeners. Hopefully that won't make us robots but will make us more discerning on what we observe and hear. On an occasion or two in a chat I picked-up on a remark in the "chat" that might have been an AI. That gave me pause, because the remarks didn't quite fit. Thanks for that caution. As always Trygve, good advice.