People stay in control
AI supports judgement and learning; it does not remove meaningful human choice or accountability.
Responsible AI & ethics
Technology should expand human agency—not replace it.
WHSF applies AI in education, careers, agriculture, information and community services. This framework sets the standard for safety, dignity, accessibility and accountable human oversight.
Six principles
Our responsible AI standard.
These principles apply to WHSF-designed tools, programme activities, partners and vendors.
Design against exclusion
Test for unequal performance and barriers affecting girls, women, people with disabilities and underserved communities.
Use the minimum data
Collect only what is necessary, explain the purpose and protect sensitive information throughout its lifecycle.
Protect children and communities
Apply safeguarding, security review, content boundaries and escalation routes before deployment.
Say when AI is involved
People should know when content, recommendations or services are AI-assisted and understand important limitations.
Monitor and correct harm
Assign owners, record decisions, review performance and provide a practical reporting and remedy process.
Ethical lifecycle
Five checks from idea to retirement.
Never acceptable
Red lines.
- Hidden manipulation or deceptive AI identity
- Automated decisions that deny essential support without human review
- Uploading confidential participant data into unapproved tools
- Biometric surveillance or profiling without a lawful, necessary and proportionate basis
- Publishing AI-generated claims as verified evidence
Report a concern
Question an AI-assisted service.
Anyone may report an error, accessibility barrier, privacy concern, suspected bias or safeguarding issue. Include the service used, what happened and the outcome you need.
Contact WHSFThis public framework should be reviewed alongside WHSF’s privacy and safeguarding policies and updated as programmes and regulations evolve.