Our Blog
/

Is this AI Slop on Your Plate This Thanksgiving?

$hero_image['alt']

Last Thanksgiving we challenged you to figure out which images were real and which were AI-generated. But over the last year, AI technology has accelerated faster than we can keep up. Tools like Sora and Veo are now used to create videos and images that are nearly indistinguishable from reality. Filling our social media feeds with “AI slop”, these tools are being used to blur the lines between reality and fiction.

AI-generation tools aren’t just the purveyors of harmless entertainment anymore; they are now being misused to generate more harmful imagery. One of the many horrors of the misuse of AI is generating deepfakes of nonconsensual intimate images, like in the case of Megan Thee Stallion’s ongoing defamation case, centered on the spread of a deepfake video she claims has damaged her reputation and mental health.

Detection of AI-generated content has become more difficult. That is why AI developers and deployers need to mitigate risks of harm and misuse. PAI’s Synthetic Media Framework was created to help address this challenge in AI. The Framework, built with industry, civil society, media/journalism, and academia partners, focuses on specific actions AI developers and distributors can take such as direct and indirect disclosure, receiving informed consent, and avoiding the distribution of unattributed content.

While regulatory bodies and regular people try to catch up to the impacts of AI-generated content, the tools to create it are still imperfect. If you know what to look out for you can still spot what is likely an AI-generated image or video.

Can you separate the truth from the “slop”?

Test your skills below. Click the video you think is AI generated in each pair.

REAL

AI

AI

REAL

REAL

AI

AI

REAL

REAL

AI

How did you do? Could you spot the AI videos above? If not here are some tips and tricks for how to spot AI content:

  1. Unusual Cropping or Angles
    • AI generated videos can oftentimes include weird angles or cropping that doesn’t make sense. If the video looks like it was filmed on a smartphone but shows black borders on the top and bottom, it is likely AI-generated.

  2. Watermarks
    • Many AI tools like Veo and Sora include watermarks in their generated content. In fact, this one of PAI’s recommendations in the Synthetic Media Framework. Did you find the watermarks in AI videos above?

  3. Anatomy and physics that doesn’t make sense
    • If items are there one second and disappear the next, or if people have more than one hand, that’s a pretty clear sign that a video was AI-generated. Did you notice the turkey carver has more than one hand?

  4. Smooth movements
    • Sometimes everything looks right! Maybe even too right. Very smooth and fluid movements can be a sign that a video is AI-generated. If those rigid and awkward movements we expect from people are not in sight, it might be AI! Notice how the pie dough is rolled out very smoothly.

  5. Poor Quality
    • This may not always be the case but AI-generated videos don’t always capture minute details. If a video is low quality or the details are indiscernible it might be AI.

  6. Lighting
    • Cinematic lighting with an iPhone or Android is pretty unusual. If the lighting looks off or isn’t hitting reflective surfaces like it should it is probably AI.