Our commitment to responsible AI

The five OECD AI principles & how we apply them

1. 1 Inclusive growth, sustainable development & well-being

   OECD: AI should benefit people and the planet by driving inclusive growth, sustainable development and well-being for all — including future generations.

   How we implement this

   • Our library spans 65 professional fields and 21 languages, making expert AI agents accessible regardless of background, geography, or specialism.
   • Agents that concentrate benefit on a narrow group or actively harm communities are rejected from the library.
   • Agents are screened with their real-world impact in mind — they should leave users better off, not worse.
2. 2 Human-centred values & fairness

   OECD: AI actors must respect the rule of law, human rights and democratic values throughout the AI lifecycle — including freedom, dignity, autonomy, privacy, non-discrimination, equality, and social justice.

   How we implement this

   • Agents are screened automatically for discriminatory language, harmful stereotypes, and content that violates individual rights before publication.
   • Every agent is framed as a tool that augments human judgment — none are positioned as replacements for human decision-making in high-stakes contexts (medical, legal, financial, safety).
   • Agents that harvest, profile, or manipulate users are categorically rejected.
3. 3 Transparency & explainability

   OECD: AI actors should commit to transparency and responsible disclosure — making AI systems understandable and their outcomes legible to the people they affect.

   How we implement this

   • Every agent displays its full system prompt: role, context, task, constraints, and output format. There are no black-box instructions hidden from users.
   • Placeholder fields ([like these]) make the agent's configurable assumptions explicit — users know exactly what they're customising.
   • Authors are credited publicly; anonymous submissions are marked as such.
4. 4 Robustness, security & safety

   OECD: AI systems should be robust, secure and safe throughout their lifecycle, with risks continually assessed and mitigated — including resilience to both accidental and deliberate misuse.

   How we implement this

   • Agents that generate harmful, misleading, manipulative, or illegal content are rejected regardless of how they're framed.
   • Contributors are encouraged to include explicit safety constraints and output guardrails within the agent's instruction set.
   • The community can flag any published agent; flagged agents can be removed if they violate these standards.
   • We do not publish agents designed for surveillance, social engineering, or evasion of security controls.
5. 5 Accountability

   OECD: AI actors should be accountable for the proper functioning of AI systems and for upholding these principles, consistent with their roles and the state of the art.

   How we implement this

   • Contributors are identified by name (or acknowledged as Anonymous) — authorship creates a line of accountability for each agent.
   • The Prompteneering team is accountable for curation decisions. Submissions are screened automatically before joining the library.
   • Agents later found to violate these principles can be removed; authors are notified where possible.
   • We publish this page as a public commitment, not a policy buried in legal text.

Read the full OECD AI Principles at oecd.ai/en/ai-principles.