Microsoft has confirmed a ban on the use of facial recognition technology by US law enforcement through its Azure OpenAI Service.
This stipulation appears in the updated service terms. According to these, Microsoft prohibits integrations with Azure OpenAI Service by “police departments or for them” for facial recognition within the United States. This applies to both current and potential future OpenAI image analysis models.
A separate new clause extends to “any law enforcement agencies globally.” It explicitly bans the use of “real-time facial recognition technology” on mobile cameras such as body-worn or vehicle-mounted ones for attempting to identify individuals in “uncontrolled, field” conditions.
The changes followed a week after Axon, a producer of law enforcement technologies, announced a product based on OpenAI’s GPT-4.
It is unclear whether Axon used GPT-4 through Azure OpenAI Service. If so, Microsoft’s updated policy might be a response to the product launch.
Microsoft Leaves Room for Flexibility
The full ban applies only to American police. It also does not extend to facial recognition using stationary cameras in controlled environments, such as in an office (although any use by US police is prohibited).
This aligns with the recent approach of Microsoft and OpenAI towards AI contracts related to law enforcement. Previously, the startup collaborated with the Pentagon on several projects, and the tech giant offered the DALL-E tool for the US Department of Defense’s military needs.
The Azure OpenAI Service became available in Microsoft Azure Government in February with additional features for government agencies and law enforcement. Microsoft Federal Vice President Candice Ling promised that the service would be “submitted for approval” to the Department of Defense to support its missions.
In March 2023, it was revealed that the FBI had been testing facial recognition on Americans for years.