Artificial intelligence policy for dry.net.au/

dry.net.au/ serves as a place of remembrance, heritage learning, and respectful storytelling. This policy ensures any use of artificial intelligence (AI) on the site honors people, place, and culture; protects privacy; and upholds accuracy, transparency, and care.

Purpose and scope

• Purpose: Define how AI is used to support research, education, and storytelling without causing harm, misrepresentation, or cultural insensitivity.
• Scope: Applies to all AI-assisted content, features, tools, workflows, and partnerships across dry.net.au/, including text, images, audio, video, data processing, and moderation.
• Audience: Contributors, moderators, partners, contractors, and visitors interacting with AI-enabled features.

Guiding principles

• Respect: Uphold dignity of deceased individuals, their families, communities, and custodians of heritage sites; never sensationalize grief or memorial content.
• Cultural care: Follow guidance from Traditional Owners and custodians; avoid publishing or generating content that risks disrespect to Aboriginal and Torres Strait Islander cultures, sacred sites, or ancestral remains.
• Truthfulness: AI outputs must be fact-checked, attributed, and corrected when errors are found; speculation is labelled clearly.
• Transparency: Disclose when content is AI-assisted and provide a clear pathway to ask questions or request corrections.
• Consent: Obtain documented permission for sensitive, personal, or identifiable memorial content wherever practicable.
• Safety: Prohibit uses that could cause physical, emotional, or financial harm; prioritize the wellbeing of visitors and communities.
• Privacy: Minimize data collection and processing; do not use personal data for model training without explicit consent.
• Accessibility: Ensure AI outputs support inclusive, plain-language communication and accessible formats.

Acceptable use and prohibited activities

• Allowed, with oversight:
  • Research assistance: Summarizing public, verifiable sources to support historical context, with citations and human review.
  • Editorial support: Grammar, clarity, and structure improvements, preserving original meaning and voice.
  • Educational content: Generating neutral, respectful explanations of heritage practices and cemetery history.
  • Moderation aid: Flagging potentially harmful or sensitive content for human review (no automated removal without oversight).
• Strictly prohibited:
  • Synthetic memorials: Fabricating people, gravestones, or events; using AI to invent or alter historical records.
  • Deepfakes and impersonation: Creating or distributing synthetic media that imitates real individuals.
  • Sensitive scraping: Harvesting obituaries, memorial posts, or personal data without explicit permission.
  • Targeting vulnerable groups: Using AI to profile or market to grieving individuals or vulnerable communities.
  • Graphic or sensational content: Generating or amplifying violence, gore, or exploitative depictions of death.
  • Cultural disrespect: Publishing AI content that reveals restricted information, misuses imagery, or disregards protocols.
  • Legal violations: Infringing copyright, privacy, consumer law, or site terms.

Content standards and editorial controls

• Disclosure: Clearly mark AI-assisted content with a notice (e.g., “This article was assisted by AI and reviewed by our editorial team.”).
• Human review: All AI-generated or AI-assisted content is edited and approved by a qualified human editor before publication.
• Accuracy and sources: Provide citations to primary or reputable sources; label uncertain information as “unconfirmed” or “in dispute.”
• Image and audio standards: No synthetic reconstructions of real individuals; any illustrative AI media must be generic, educational, and clearly labelled as synthetic.
• Cultural protocols: Consult with relevant custodians for content involving sacred sites, ancestral remains, or restricted knowledge; respect requests for removal or revision.
• Children and families: Extra caution for content involving minors or active grief; avoid intrusive detail and identifiable information without consent.
• Right to request changes: Families, custodians, and community representatives can request corrections, context additions, or removals; these are prioritized and documented.

Data, privacy, and model governance

• Data minimization: Collect only what is necessary for site function; do not retain sensitive data longer than needed.
• No training on user submissions: User-submitted memorials, comments, images, or messages are not used to train AI models unless explicit, informed consent is provided.
• Security: Apply reasonable safeguards to prevent unauthorized access or misuse of data processed by AI systems.
• Third parties: Any vendor or partner must meet or exceed this policy; agreements specify privacy, security, and cultural respect standards.
• Updates: Disclose substantive changes to AI features or data practices; maintain a version history of this policy.

Incident response and accountability

• Reporting: Provide an accessible contact for concerns about AI content, cultural protocols, privacy, or accuracy: contact@zazu.com.au (placeholder).
• Triage: Acknowledge within a reasonable timeframe; prioritize issues involving harm, cultural sensitivity, or privacy.
• Remediation: Correct, annotate, or remove content as needed; document the change and notify affected parties where appropriate.
• Audit and review: Conduct regular audits of AI-assisted workflows for bias, accuracy, and compliance; publish a brief annual summary of findings and improvements.

Community engagement and continuous improvement

• Consultation: Engage Traditional Owners, local historians, heritage organizations, and community groups in content planning and review where relevant.
• Feedback loops: Invite public input on AI disclosures, content standards, and cultural protocols; incorporate learnings into updates.
• Education: Offer guidance on memorial etiquette, respectful photography, and heritage literacy; clarify how AI can assist—without replacing human care.

Appendices

AI disclosure template

• Notice: “Parts of this page were assisted by AI to improve clarity and structure. All facts were checked by our editorial team.”
• Contact: “Questions or concerns? Email contact@zazu.com.au.”

Consent checklist for sensitive memorial content

• Identity: Confirm relationship or custodianship.
• Scope: Define what can be shared, how, and for how long.
• Withdrawal: Provide a simple process to revoke consent.
• Cultural protocols: Note any restrictions or required context.

Editorial review checklist for AI-assisted pieces

• Accuracy: Verify names, dates, locations, and citations.
• Tone: Ensure language is respectful and non-sensational.
• Cultural care: Confirm compliance with protocols and permissions.
• Privacy: Remove unnecessary or identifying details.
• Labeling: Add AI-assistance disclosure where applicable.

Risk matrix (summary)

• Low risk: Generic educational explanations with sources and human review.
• Medium risk: Site descriptions with community context; requires local consultation.
• High risk: Content involving specific individuals, gravesites, or restricted cultural information; requires explicit consent and custodian approval.

Policy ownership and updates

• Owner: dry.net.au/ Editorial & Community Care Team.
• Effective date: 30 September 2025
• Review cycle: Reviewed at least annually, or sooner if AI features change, new risks emerge, or community guidance is updated.

If you have questions or concerns about our AI use, please contact:
📧 Contact Us

📍 dry.net.au/ – Sutherland Shire, NSW, Australia