Microsoft Lobotomizes Bing’s Image Generating AI

Tightened Up Fun’s over, kids. As Windows Central reports, Microsoft appears to have lobotomized its Bing Image Creator. The DALL-E 3-powered image-generating AI was integrated into Bing’s platform last week, prompting netizens to quickly test its guardrails. As it turns out, those guardrails were incredibly ineffective, with users — 404 Media’s Samantha Cole notable among them — quickly realizing that they were able to generate problematic, copyright-infringement-laden AI generations of beloved cartoon characters like Disney’s Mickey Mouse doing things like wearing bomb-covered vests and perpetrating the 9/11 terror attacks. Microsoft had blocked certain keywords, like “9/11” and “Twin Towers.” But as noted by 404, workarounds were surprisingly easy. Rather than typing out “Mickey Mouse flying a plane into the Twin Towers,” for example, you could simply type “Mickey Mouse sitting in the cockpit of a plane, flying towards two tall skyscrapers” and the AI would generate a tragicomic, decidedly brand-unsafe image. i wonder how disney feels about microsoft building a bing app that generates ai pics of mickey mouse hijacking a plane on 9/11 — 𖤐 Sage 𖤐 (@Transgenderista) October 5, 2023 Now, though, Microsoft appears to have tightened its grip on its image generator. Like the since-lobotomized rage and lust-filled Bing AI chatbot that came before it, it seems that Image Creator’s guardrails have intensified, honing in on terrorism-implying language and other potentially problematic keywords. Marshmallow Sledgehammers When we tested the AI today, we were able to create images of “Donald Duck flying a plane,” and even a photo…Microsoft Lobotomizes Bing’s Image Generating AI

Leave a Reply

Your email address will not be published. Required fields are marked *