Microsoft restricts access to some of its artificial intelligence services

Artificial intelligence holds the future of technology, but in the wrong hands, it can cause trouble for many people. That's why Microsoft announced some changes to its responsible AI standards, introducing a new limited access policy. The company said in a number of posts on June 21 that it would now remove and restrict access to some of its services that use AI. In particular, these policy changes will affect Microsoft's Azure Face facial recognition service and custom neural voices.

One of the features that will be greatly affected by this change is the controversial facial analysis technology, which aims to infer an individual's emotional state and identify different human attributes such as gender, age, smile, Facial hair, hair and makeup. Microsoft said the move follows some privacy concerns and a lack of scientific consensus on the definition of the concept of "emotion."

“We work with internal and external researchers to understand the limitations and potential benefits of this technology and weigh in on the trade-offs,” said Sarah Bird, principal group product manager for Azure AI. "Particularly in the case of emotion classification, these efforts raise important questions about privacy, a lack of consensus on the definition of "emotion," and an inability to generalize between facial expressions and emotional states across use cases, regions, and demographics. Contact. API access to the ability to predict sensitive attributes also opens up a variety of ways it could be abused — including subjecting people to stereotypes, discrimination, or unfair denial of service."

As a result, the company announced that it has The attribute detection function will be available to new customers on June 21st, and will be discontinued for existing customers on the 30th of the same month. Still, Microsoft says it "recognizes that these features can be valuable when used in a controlled set of accessibility scenarios." Therefore, there is an exception to the ongoing provision of these features in apps designed for people with disabilities, e.g. Seeing AI.

On the other hand, Microsoft will restrict applicants from accessing facial recognition services in the Azure Face API, Computer Vision, and Video Indexer. According to the limited access policy introduced by Microsoft, customers must pass use cases and customer qualification requirements to access operations. Current customers will have one year, until June 30, 2023, to apply for and be approved to continue using facial recognition services. Meanwhile, additional face detection features for detecting blur, exposure, glasses, head pose, landmarks, noise, occlusions, and facial bounding boxes will remain available without apps.

Finally, Microsoft's custom neural speech feature will also face some restrictions to prevent possible abuse, according to Bird. "Based on what we learn from custom neural speech, we will apply similar controls to our facial recognition service," Chief Responsible Artificial Intelligence Officer Natasha Crampton wrote in a separate blog post. "Following a transition period for existing customers, we are limiting access to these services to managed customers and partners, narrowing use cases to predefined acceptable use cases and leveraging technical controls designed into the services."

The above is the detailed content of Microsoft restricts access to some of its artificial intelligence services. For more information, please follow other related articles on the PHP Chinese website!