It seems like every day there's a new concern over AI, despite plenty of major brands jumping into the trend in the hopes of being the one to create the next big AI innovation. However, a long-term AI engineer at Microsoft has wrote a letter to the FTC, voicing concerns about the Copilot Designer tool creating some disturbing images.
Microsoft has made it's own AI a huge part of Windows 11, either forcing users to remove Copilot from Windows 11 or delighting people and getting them to subscribe to Copilot Pro. However, Copilot Pro is proving to be falling into similar traps as Google's Gemini, with one employee reporting that Copilot Designer is generating harmful images.
In a report from CNBC, AI Engineer Shane Jones, who has worked at Microsoft for six years, has apparently wrote a letter to the FTC, sharing concerns about the AI image generator from Microsoft. While not working on Copilot in a professional capacity, Jones is one of many employees who tests Copilot (in his free time, may I add), in order to find problems that the Windows AI could be creating.
In his letter to the FTC and the Microsoft board, Jones expressed concerns with Copilot Designer, which has reportedly been generating images of demons and monsters linked to abortion rights terminology, underage drinking and drug use, teenagers with assault rifles, and more. In our testing, we weren't able to recreate these images, but CNBC has apparently been able to create similar images just this week.
Jones has been reporting these findings since December 2023, and while Microsoft has acknowledged his concerns, the company isn't taking the tool offline until things are improved. He reportedly wrote a statement on LinkedIn, which Microsoft prompted him to remove, forcing him to write letters to the FTC, the board, and US Senators.
This comes shortly after Google's Gemini AI tool created indecent images and questionable responses, forcing Google to take the AI offline. However, Copilot Designer is still available as I write this article.
Hopefully, these safety concerns will be taken seriously, and force Microsoft to take the image generator offline until it can take these concerns seriously. Similar concerns have been raised about the potential harm that the upcoming text-to-video generator from OpenAI, Sora, but we have yet to see any actually harmful results, since the OpenAI Sora release date hasn't been revealed yet.
Explore new topics and discover content that's right for you!