Human rights in the age of

HUMAN RIGHTS IN THE AGE OF ARTIFICIAL INTELLIGENCE
The Right to an Explanation provides for a person to get an explanation about how an automated decision 
is made pertaining to that person. This right ensures entities understand how the systems they use actually 
work, and it pushes AI developers to continue working to make AI understandable.
The Right to Object gives people the ability to contest most processing of their personal data by an entity 
when the data are used for direct marketing, automated decision making (where no human intervention 
will take place), research and statistics, or for an entity’s “legitimate interest.” This right allows for direct 
challenges to decisions made using AI systems. It is particularly important for government use of AI in ways 
that can be discriminatory. It also ensures there is a human in the loop in important automated decision-
making systems, which adds a layer of accountability.
AI-SPECIFIC RECOMMENDATIONS FOR GOVERNMENT AND THE PRIVATE SECTOR 
Because AI is a diverse field, the potential for interference with human rights depends both on the type of 
data a system uses and the context in which the system is implemented. For example, there are fewer and 
different human rights risks posed by a city government’s use of AI to optimize water usage than a police 
department’s use of a criminal risk assessment tool. With this in mind, we recommend different approaches 
for government and the private sector. 
Recommendations for government use of AI 
AI systems for government often implicate value judgments that are necessarily linked to the political process 
in free and democratic government systems. For this reason, and because governments can directly deprive 
people of their liberty, this report recommends higher standards for the public sector regarding the use of 
AI. States bear the primary duty to promote, protect, respect, and fulfill human rights under international 
law, and must not engage in or support practices that violate rights, whether in designing or implementing AI 
systems. They are required to protect people against human rights abuses, as well as to take positive action 
to facilitate the enjoyment of rights.
116
The recommendations below articulate a framework for government 
decision making in general, not just in AI. They apply to any type of algorithmic decision making system, 
regardless of whether it uses AI.