| AI systems should treat all individuals and groups impartially, equitable and free from prejudice. Regulators should assess systems for potential bias and ensure transparency around decisions. | | Regulators should encourage system developers to ensure people understand AI decisions by providing clear explanations. Systems should be transparent about development, capabilities, limitations and real-world performance. | | Regulators should ensure that AI is human-centric, with appropriate oversight and controls. Organisations remain responsible for systems and their impacts. Proportionate governance measures should manage risks. | | AI systems should reliably behave as intended while minimising unintentional and potentially harmful consequences. Threats like hacking should be anticipated and systems designed accordingly. | | The privacy rights of individuals, groups and communities should be respected through adequate data governance. The collection, use, storage and sharing of data must be handled sensitively and lawfully. |
|---|