Tips for spotting AI-generated election disinformation and propaganda

  With less than six months to go until we elect the next U.S. president, voters must be savvy to spot disinformation. Artificial intelligence tools are making that task more difficult by allowing people to easily generate fake images and video intended to sway voters.

  Here are some questions taken from various news articles about this technology to help you become an AI watchdog this election season. However, be forewarned: The Brennan Center for Justice believes AI technology is improving at such a pace that looking for such clues may be a fruitless effort. Nevertheless, if you answer “yes” to any of the questions below, you may be looking at AI-generated video and photos.

• As subjects move their heads, do their faces change in abnormal ways?
• Is the subject’s skin too smooth or too wrinkled?
• Are there shadows in strange places?
• Are the subject’s fingers or limbs missing? Are there too many fingers and limbs? Does the subject’s clothing seem misaligned?
• Does the subject’s facial hair look unrealistic or otherwise “wrong”?
• Do the subjects have strange blinking patterns?
• Are the subject’s mouth movements out of sync with the audio?
• Is the subject’s chin moving unnaturally?
• Is the background in the image or video unrealistic or just “off”?
• Are objects in the frame not proportional?

The bigger picture

  Of course, not all election disinformation uses AI-generated images and video. Here’s how to discern if what you are reading is true, according to the Brennan Center for Justice and other sources.

• Use fact-checking sites like PolitiFact, AP Fact Check, Reuters Fact Check, and Snopes to find out if what you’re reading is the real deal.
• Seek out authoritative context. Are trustworthy sources named or referenced as part of the information?
• Look for the original source of content. Where did the post come from? Is the source reliable? Here’s a list of news sources Americans trust most.
• Be wary of highly emotional content. It can be used to impair judgment to deliver false information more easily.
• When seeking voting information, use official government election resources. Also, check out the Southern Poverty Law Center’s guide to voting in the South.
• Pay attention when using search engines. Many search engines, including Google, use generative AI to summarize search results at the top of their findings. Much like ChatGPT and other generative AI models, which should not be used to obtain election information, the accuracy of these summaries cannot always be trusted.

  During this moment when election disinformation poses a grave threat to our elections and the democratic process, we must remain vigilant. By staying informed, exercising critical thinking, and using reliable sources, we can protect the integrity of our voting system and uphold our democratic values.

  This article was published by the Southern Poverty Law Center, an Alabama-based civil rights organization.