Ethics guidelines for trustworthy AI
European Union · Guideline · High-Level Expert Group on AI
Our summary
The High-Level Expert Group on AI released guidelines setting out principles for artificial intelligence systems that are lawful, ethical, and robust. The guidelines identify seven key requirements that trustworthy AI systems should meet: human agency and oversight, technical robustness and safety, privacy and data governance, transparency, diversity and fairness, societal and environmental well-being, and accountability. The guidelines include an assessment list to help organizations verify their compliance with these requirements.
Written by the Index in our own words. It is not legal advice; the official text is the authority.
Inside this policy
Every section of this policy in order. Our notes summarise each one in plain language and do not quote the official text, which is the authority.
Drafted by the Index An automated summary of the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
4 sections
Jump to a chapter
Download the Guidelines in your language below
Establishes seven key requirements that AI systems must satisfy to be considered trustworthy: human agency and oversight through human-in-the-loop mechanisms, technical robustness with safety fallbacks, privacy and data governance, transparency with explainability adapted to stakeholders, diversity and non-discrimination to avoid unfair bias, societal and environmental well-being considering future generations and other living beings, and accountability with auditability and redress mechanisms. An assessment list is provided to verify compliance with each requirement.
Download the Definition of AI in your language below
Piloting Process
Starting from June 26th, the assessment list was tested through a public piloting process inviting all stakeholders to provide feedback on its practical implementation. Stakeholders submitted input through open surveys, in-depth interviews with representative organisations across sectors, and continuous uploads of feedback and best practices via the European AI Alliance, with the piloting phase ending on December 1st, 2019.
The piloting phase closed on 1 December 2019
Following stakeholder feedback from the piloting phase, the AI HLEG released the final Assessment List for Trustworthy AI in July 2020 as a practical self-assessment tool for developers and deployers to implement the key requirements. The tool is available as a prototype web-based platform and in PDF format.
Controls
What this policy requires, forbids, grants or penalises, one control at a time and in our own words. Each control links to the official text, which is the authority. This is not legal advice.
Drafted by the Index Drafted automatically from the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
Themes covered 6 of 35
Choose a theme to see only its controls. About the themes
Risk and scope 0 of 4
- Risk classification, not covered
- Prohibited practices, not covered
- Risk management system, not covered
- Impact assessment, not covered
Data and privacy 2 of 4
- Cross-border data transfer, not covered
- Copyright and training data, not covered
Transparency and records 1 of 4
- Labelling of synthetic content, not covered
- Technical documentation, not covered
- Record keeping and logging, not covered
Technical safeguards 2 of 4
- Bias and non-discrimination, not covered
- Testing and evaluation, not covered
Assurance and lifecycle 1 of 4
- Registration and notification, not covered
- Incident and breach reporting, not covered
- Post-market monitoring and corrective action, not covered
Rights and protected groups 0 of 4
- Automated decision-making rights, not covered
- Children and vulnerable groups, not covered
- Biometric identification and surveillance, not covered
- Whistleblowing, complaints and redress, not covered
Roles and organisation 0 of 6
- Supply chain and provider-deployer duties, not covered
- General purpose and foundation models, not covered
- Accountability and governance roles, not covered
- Training and AI literacy, not covered
- Public sector use and procurement, not covered
- Innovation support and sandboxes, not covered
Enforcement and sectors 0 of 5
- Enforcement, penalties and regulators, not covered
- Sector rules: health, not covered
- Sector rules: finance and insurance, not covered
- Sector rules: employment, not covered
- Sector rules: education, not covered
The full list of controls could not be loaded, so only the first 7 are shown.
Download the Guidelines in your language below 6 controls
- ObligationApplies to: ProviderDrafted by the Index
AI systems should enable people to make informed decisions and safeguard their fundamental rights.
- ObligationApplies to: ProviderDrafted by the Index
AI systems should be resilient, secure, and include fallback plans to prevent unintended harm.
- ObligationApplies to: ProviderDrafted by the Index
AI systems should be accurate, reliable and reproducible.
- ObligationApplies to: ProviderDrafted by the Index
Providers should ensure full respect for privacy and data protection with adequate data governance mechanisms.
- ObligationApplies to: ProviderDrafted by the Index
Providers should ensure data quality and integrity, and legitimise access to data.
- ObligationApplies to: ProviderDrafted by the Index
AI systems should present transparent data, system and business models, with traceability mechanisms and decisions explained in ways suited to the relevant audience.
The piloting phase closed on 1 December 2019 1 control
- ObligationApplies to: ProviderDrafted by the Index
Developers and deployers should use the Assessment List for Trustworthy AI as a practical self-assessment tool to implement the seven key requirements.
Change history
No changes recorded since we began monitoring this policy.