How to Avoid Falling for Fake Videos in the Age of AI Slop
Why fake videos spread so easily, and what actually helps people spot them.
This piece was co-written by Dr. Beth Malow and is also cross-posted here.
We’re entering an era of what’s often called AI slop: an endless stream of synthetic images, videos, and stories produced quickly, cheaply, and at scale. Sometimes these fake videos are harmless or silly, like 1001 cats waking up a human from sleep. Other times, they are deliberately designed to provoke outrage, manipulate identity, or push propaganda.
Twitter/X is flooded with this content, but it has spread well beyond it. You can now find AI-generated misinformation on platforms like LinkedIn (a place many people associate with professionalism and credibility). I recently debunked a fake AI video there, and unfortunately, this kind of content is becoming more common across nearly every major social media platform.
This is a serious problem. But it’s not a hopeless one.
To navigate this new information environment, we need to combine psychological literacy, media literacy, and policy-level change. Here’s how:
1) Understand Our Own Psychological Biases (Psychological Literacy)
Think about how you feel when you come across a post that taps into strong emotions around a major cultural or political issue. Highly charged topics can provoke immediate reactions across the political spectrum. Because these issues are so emotionally loaded, they are especially vulnerable to being exploited through misleading or fabricated videos designed to spark outrage, regardless of where someone falls on the political aisle. The psychology behind falling for AI-generated misinformation isn’t fundamentally new. The process is largely the same as with other forms of misinformation, but AI takes it to a whole new level– it dramatically lowers the cost and effort required to produce and spread it at scale.
These videos are engineered to hit multiple emotional and cognitive triggers at once:
authority, danger, injustice, resistance, and moral certainty. They often present a clean, cinematic moment of confrontation. You don’t need backstory. You don’t need names. You’re meant to feel first and share second.
This effect is especially powerful when the content aligns with your identity.
As Matthew has written before about fake AI accounts, people are motivated to believe what fits their values, grievances, and group identities, not necessarily what’s true. When a video confirms what you already believe about politics, culture, or power, authenticity becomes secondary. What matters is that it feels right.
AI makes this dynamic worse. No real people need to be involved. No real event has to occur. There may be nothing to fact-check at the source, because the source itself is fictional.
That’s why emotional intensity is one of the most important early warning signs.
When something instantly makes you angry, vindicated, or morally superior, that’s often your cue to pause, not because the reaction is wrong, but because it’s likely being exploited.
The goal isn’t to suppress emotion. It’s to recognize when emotion is being used as a shortcut around verification, and being used to manipulate you. This is what Matthew has been calling psychological literacy, and it’s one of the most effective defenses we have against modern misinformation.
2) Lateral Reading Is Still the Best Tool We Have (Media Literacy)
When people try to fact-check AI videos, their instinct is often to stare harder at the content itself such as examining faces, counting fingers, looking for visual glitches.
Sometimes that helps, but AI-generated fakes are always changing and improving.
The most effective fact-checking strategy we have isn’t vertical reading (scrutinizing the video itself). It’s lateral reading: leaving the content entirely to verify it elsewhere.
Professional fact-checkers do this instinctively. Instead of asking, “Does this look real?” they ask:
Are credible news outlets reporting this incident?
Does the video trace back to a legitimate source or named reporter?
When and where was it first posted?
Is there independent confirmation outside social media?
In practice, that means opening new tabs and asking:
Who shared this?
What’s their source?
Can I find confirmation beyond this platform?
In many AI misinformation cases, the answers converge quickly. No verified outlets report the event. The account has no meaningful history outside viral clips. The video circulates primarily within a single platform or community.
You can still use visual cues, awkward movement, odd writing/symbols, overly smooth lighting, lifeless background characters, but these should be treated as secondary clues, not primary evidence. There are some digital tools that can help, such as Google’s reverse image search, and look for Matthew’s future post reviewing more of these digital tools.
If you want a simple framework, try the SIFT method:
Stop before reacting (this includes not reposting on social media until you are certain of the credibility of what you are seeing)
Investigate the source (have a low threshold to open new tabs)
Find better coverage (have major news outlets verified the claim?)
Trace claims back to their origin (this is where reverse image search can help)
Lateral reading works because it doesn’t require you to outsmart AI. It only requires you to check whether reality agrees.
3) Policy Changes and Platform Accountability
Individual skills matter. Community norms matter. But at this point, policy intervention is likely required.
Social media platforms are not optimized for truth, they’re optimized for engagement. AI slop thrives precisely because it hijacks attention cheaply and at scale. Synthetic outrage is profitable, even when it’s socially corrosive.
We need stronger expectations around:
Clear labeling of AI-generated content
Transparency about synthetic accounts
Accountability for repeat offenders
Demonetization of deceptive AI slop
Transparency and accountability are essential for breaking echo chambers at scale. The same logic applies here. YouTube is making a priority to address AI slop on their website, and other social media platforms are giving users more control over what they see, but we’d like to see this happening more broadly.
AI slop isn’t helping society, even if it helps pad quarterly earnings of those stirring up conflict or outage (see post on conflict entrepreneurs) by hijacking our attention.
Conclusion
The most dangerous thing about fake AI videos isn’t that people believe them once. It’s that repeated exposure erodes trust altogether: in media, in institutions, and eventually in one another.
Defending against this doesn’t require becoming a forensic AI expert. It requires something harder but more durable: understanding our own biases, verifying claims laterally, and demanding accountability from systems designed to exploit attention.
As AI becomes a bigger part of how reality is represented to us, psychological literacy is a prerequisite for staying oriented in the world.
And that, more than any single tool or policy, is how we keep from getting lost in the slop.
 | A guest post by
|