Microsoft has amended the terms of service for Azure OpenAI Service, which allows professionals to access OpenAI language models, to more explicitly prohibit their use by US police.
OpenAI partner, Microsoft does not want the American police to use its artificial intelligence (AI) service for facial recognition. As reported by the specialist site TechCrunch, the company reaffirmed this ban by modifying the general conditions of use of Azure OpenAI Service.
This service allows businesses and developers to access OpenAI’s language models, such as GPT-4. Its terms of service now state that “integrations with Azure OpenAI Service must not be used for facial recognition purposes by or for any law enforcement agency in the United States.”
Formal ban
Microsoft also specifies that law enforcement is not authorized to use these integrations for “any real-time facial recognition technology on mobile cameras (…) to attempt to identify individuals in uncontrolled environments” .
Concretely, police officers on patrol wearing a camera using facial recognition for this purpose cannot use Azure OpenAI Service. Likewise when these cameras are mounted on the dashboard of their car.
This rule change comes a week after Axon, a maker of law enforcement technology and weapons products, announced a new product using GPT-4. Called Draft One, it is an AI capable of generating police reports from body camera recordings.
Axon was able to use OpenAI’s technology after the company changed its terms of service last January, no longer explicitly prohibiting the use of its AI model for military purposes. Microsoft does not, however, specify whether it was Axon’s announcement that pushed it to change its rules.
