AI used to analyze more than 2000 posts from 25 brands

WBUR 90.9 FM Radio – Air Date: September 17, 2024
Reporter: Amy Sokolow, Associate Producer
News Anchor, Steve Brown

Audio and transcription below:

Brown: Boston University found the vast majority of Instagram posts by some e-cigarette and vape brands did not include health warnings about the products. The posts exploited a loophole to evade federal regulations on advertising and targeting young people. Researchers used artificial intelligence to analyze more than 2000 posts from 25 brands. BU Media Science Professor Tracy Hong says many of the companies added deceptive warning labels.

Hong: “For example, one warning label that was very common was ‘this product contains non-tobacco, nicotine, Nicotine is an addictive product. This product is not derived from tobacco.’ To me, it doesn’t even read like a warning label. It suggests as if the product is safer.”

Brown: She says that the AI model found that those posts with accurate health warnings got less social media engagement, reducing their online reach.”

Learn more about this research here.