Staff Reports
We’ve seen this movie before: a foreign influence operation sneaks into our social feeds, pretends to be one of us, and pokes at whatever raw nerve is already exposed. Only now the puppeteers have better tools — generative AI, polished imagery, and an old playbook given a new coat of paint. Researchers warn the same China‑linked network analysts have tracked for years — Spamouflage, Dragonbridge and their cousins — is back and sharper than it looks.
Old operation, new tricks
This isn’t a spontaneous troll farm. It’s a long‑running, China‑linked influence ecosystem that’s evolved since researchers first spotted it years ago. Now investigators from groups like Graphika and the DFRLab say the operation is using AI‑generated imagery and fake event flyers to try to disrupt critics of Beijing — and to make local fights look like national crises.
That shift matters. AI makes thousands of believable-looking assets in minutes. A disinformation outfit with a keyboard and a budget can replicate local voices, seed divisive content across Facebook, X and Instagram, and then watch whichever side of the fight bites first. Even when most posts flop, one well-placed flyer or viral clip can change a meeting turnout, scare an organizer, or mainstream a lie.
Real people, real consequences
The targets aren’t shadowy think tanks — they’re exiled Tibetan voters, overseas activists, and community organizers trying to hold a town hall without chaos. In one documented case, coordinated Facebook and Instagram clusters used AI art and fake pages to meddle in a Tibetan parliamentary election abroad; in another, manipulated event flyers aimed to disrupt anti‑CCP gatherings in Europe and the U.S. Those are small groups, but the cost is immediate: canceled events, extra security, and people who stop showing up because they don’t trust what they see online.
Think of a local veterans’ group or a school board meeting — the same tools can be turned on any corner of civic life. That erosion of trust is a tax on civic engagement, and it’s paid by ordinary folks who just want to live and argue in peace without foreign bots picking sides.
Platforms: takedowns aren’t a cure
Meta, Google and others have removed six‑figure piles of inauthentic assets this year, and that’s something. But takedowns are a bandage. The networks adapt, re‑emerge and reuse assets across platforms. The companies remove accounts; the operation tweaks tactics and pops back up. Meanwhile, transparency reports trickle out and the public moves on.
If you’re tired of corporate-speak, here’s the blunt version: Big Tech gets the problem when someone points a finger at them, acts for a week, then files a quarterly report and calls it fixed. That won’t stop a determined, well‑resourced foreign actor that can buy or build AI tools and a distribution plan.
What to do about it
There are practical steps that don’t require turning the internet into a government megaphone. Platforms must standardize disclosures about coordinated inauthentic behavior, speed up cross‑platform cooperation, and publish usable indicators so researchers and defenders can trace campaigns. Congress and national security agencies should be willing to brief the public when attribution meets a reasonable bar — not to stifle speech, but to defend our civic space.
And citizens need stronger media literacy. If you run a town meeting, check the flyer’s source. If you see an account behavior that’s robotic, report it and then tell your neighbors. Defending free speech also means defending the context in which speech actually matters.
We can laugh at late‑night panels or shrug at another takedown memo, or we can demand we stop letting foreign actors turn our arguments into theatre. Which will it be?
