F12. Algorithmic content curation, recommendation, and/or ranking systems

Companies should clearly disclose how users’ online content is curated, ranked, or recommended.

Elements:

1. Does the company clearly disclose whether it uses algorithmic systems to curate, recommend, and/or rank the content that users can access through its platform?
2. Does the company clearly disclose how the algorithmic systems are deployed to curate, recommend, and/or rank content, including the variables that influence these systems?
3. Does the company clearly disclose what options users have to control the variables that the algorithmic content curation, recommendation, and/or ranking system takes into account?
4. Does the company clearly disclose whether algorithmic systems are used to automatically curate, recommend, and/or rank content by default?
5. Does the company clearly disclose that users can opt in to automated content curation, recommendation, and/or ranking systems?

Algorithmic system — A system that uses algorithms, machine learning and/or related technologies to automate, optimize and/or personalize decision-making processes.

Clearly disclose(s) — The company presents or explains its policies or practices in its public-facing materials in a way that is easy for users to find and understand.

Curate, recommend, and/or rank — The practice of using algorithms, machine learning and other automated decision-making systems to manage, shape, and govern the flow of content and information on a platform, typically in a way that is personalized to each individual user.

Content — The information contained in wire, oral, or electronic communications (e.g., a conversation that takes place over the phone or face-to-face, the text written and transmitted in an SMS or email).

Users — Individuals who use a product or service. This includes people who post or transmit the content online as well as those who try to access or receive the content. For indicators in the freedom of expression category, this includes third-party developers who create apps that are housed or distributed through a company’s product or service.

Indicator guidance: Algorithmic content curation, recommendation, and ranking systems play a critical role in shaping what types of content and information users can see and access online. In addition, systems that are optimized for user engagement can have the effect of prioritizing controversial and inflammatory content, including content that is not protected under international human rights law. Over time, reliance on algorithmic curation and recommendation systems that are optimized for engagement can alter the news and information ecosystems of entire countries or communities. These systems can be manipulated to spread disinformation and otherwise distort the information ecosystem, which can in turn fuel human rights abuses.

Companies should therefore be transparent about their use of automated curation, recommendation, and ranking systems, including the variables that influence such systems. Companies should publish information about whether they use algorithmic content curation, recommendation, and ranking systems (Element 1); how they work (Element 2); and what options users have to control how their information is used by these systems (Element 3). Companies should further disclose whether such systems are automatically on by default (Element 4). They should also clearly disclose whether users can “opt-in” to have their content automatically curated by the algorithmic system (Element 5).

• Company human rights policy
• Company artificial intelligence policies, including AI principles, frameworks, and use guidelines
• Help pages describing how feed settings, home page settings, search results, recommendations, user interests, or topics are affected by algorithms