Microsoft bans US police from using its AI
Microsoft has officially confirmed the ban on using generative artificial intelligence for facial recognition by US police departments through its corporate Azure OpenAI Service.
On Wednesday, the company amended the service’s terms of use, which now clearly prohibit integrations with Azure OpenAI Service “for or on behalf of” US police departments for facial recognition purposes. The ban applies to integrations with current and likely future OpenAI models capable of analyzing images.
A separate clause addresses “all law enforcement agencies worldwide” and prohibits the use of “real-time facial recognition technology” on mobile cameras, such as body cameras and dashcams, to identify people in an “uncontrolled environment.”
The policy changes came a week after Axon, a manufacturer of technology products and weapons for military and law enforcement agencies, announced a new product using OpenAI’s GPT-4 generative text model to summarize audio recordings from body cameras. Critics quickly pointed out potential issues, including “hallucinations” (even the best generative AI models today invent facts) and racial biases inherited from training data, which is particularly troubling given that people of color are much more likely to be stopped by police than their white counterparts.
It’s unclear whether Axon used the GPT-4 model through Azure OpenAI Service and whether Axon’s product launch was the reason for Microsoft’s policy update. OpenAI had previously already restricted the use of its models for facial recognition through its APIs.
The new terms leave Microsoft room to maneuver. The full ban on using Azure OpenAI Service applies only to US police, not international law enforcement agencies. Additionally, it doesn’t extend to facial recognition performed with stationary cameras in a controlled environment, such as an office (although the terms prohibit any use of facial recognition by US police).
