AI services may not be used for harmful social scoring or profiling, or criminality risk assessment from profiling alone
You agree not to use the AI services for social scoring or predictive profiling that would lead to discriminatory, unfair, biased, detrimental, unfavorable or harmful treatment of persons or groups. You also agree not to use them for assessing the criminality risk of natural persons based solely on profiling or on assessing their personality traits and characteristics. That second prohibition does not apply to AI services used to support a human assessment of a person's involvement in criminal activity that is already based on objective and verifiable facts directly linked to a criminal activity.
Supporting quotations
For social scoring or predictive profiling that would lead to discriminatory, unfair, biased, detrimental, unfavorable, or harmful treatment of certain persons or groups of persons;
For the assessment of criminality risk of natural persons based solely on the profiling of a natural person or on assessing their personality traits and characteristics. This prohibition shall not apply to AI services used to support the human assessment of the involvement of a person in a criminal activity, which is already based on objective and verifiable facts directly linked to a criminal activity;
Qualifications & exceptions
The criminality-risk prohibition covers assessments based solely on profiling or personality traits and characteristics.
It does not apply to AI used to support a human assessment already based on objective, verifiable facts directly linked to criminal activity.
Inference & Profiling · Restriction or exception
AI services may not be used for biometric categorization, untargeted face scraping or police real-time face recognition
You agree not to use the AI services, based on biometric data, to categorize people or to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, or sex life or sexual orientation, except for labelling or filtering of lawfully acquired biometric datasets such as images, or categorizing biometric data in the area of law enforcement. You also agree not to use them to create or expand facial recognition databases through untargeted scraping of facial images from the internet or CCTV footage, or for any real-time facial recognition technology on mobile cameras used by any law enforcement globally to try to identify individuals in uncontrolled 'in the wild' environments, which includes without limitation police officers on patrol using body-worn or dash-mounted cameras to identify people in a database of suspects or prior inmates.
Supporting quotations
Based on their biometric data, to categorize people or to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, or sex life or sexual orientation, except for any labelling or filtering of lawfully acquired biometric datasets, such as images, based on biometric data or categorizing of biometric data in the area of law enforcement;
To create or expand facial recognition databases through the untargeted scraping of facial images from the internet or CCTV footage;
For any real-time facial recognition technology on mobile cameras used by any law enforcement globally to attempt to identify individuals in uncontrolled, "in the wild" environments
Qualifications & exceptions
The biometric categorization ban has exceptions for labelling or filtering lawfully acquired biometric datasets and for categorizing biometric data in the area of law enforcement.
The real-time facial recognition ban is about law enforcement use on mobile cameras to identify individuals in uncontrolled environments.
AI services may not be used to infer people's emotional states from physical or behavioral characteristics
You agree not to use the AI services to attempt to infer people's emotional states from their physical, physiological or behavioral characteristics, such as facial expressions, facial movements or speech patterns, including inferring emotions such as anger, disgust, happiness, sadness, surprise or fear. The passage states no exceptions.
Supporting quotations
you agree that you will not use the AI services
To attempt to infer people’s emotional states from their physical, physiological, or behavioral characteristics (e.g., facial expressions, facial movements, or speech patterns), including inferring emotions such as anger, disgust, happiness, sadness, surprise, fear, or other terms commonly used to describe a person’s emotional state.
Qualifications & exceptions
Covers inferring emotional states from physical, physiological or behavioral characteristics.
The AI usage restrictions list has nine prohibited uses, ending with inferring emotional states
The AI Services usage restrictions are a numbered list of nine uses that you agree not to make of the AI services, in addition to the Code of Conduct: unsupervised consequential decisions, deception or subliminal manipulation, exploiting vulnerabilities, social scoring or predictive profiling, criminality-risk assessment from profiling alone, biometric categorization of sensitive traits, untargeted facial-image scraping for recognition databases, real-time law-enforcement facial recognition on mobile cameras in uncontrolled environments, and, as the ninth and last, attempting to infer people's emotional states from physical, physiological or behavioral characteristics. Items 1 to 8 are described in their own cards; the list closes with item 9, which names facial expressions, facial movements and speech patterns as examples.
Supporting quotations
For any real-time facial recognition technology on mobile cameras used by any law enforcement globally to attempt to identify individuals in uncontrolled, "in the wild" environments
To attempt to infer people’s emotional states from their physical, physiological, or behavioral characteristics
Qualifications & exceptions
Item 9 prohibits attempting to infer emotional states and lists its examples with 'e.g.', so they are illustrative.