Question 1

What is Edyant?

Accepted Answer

Edyant is an AI ethics and governance organisation that benchmarks large language models across 15 ethical dimensions and publishes GRACE — the Global Repository of AI Consensus at Edyant — a living, versioned AI governance standard synthesised from academic, policy, and industry contributions worldwide. Edyant partners with academia, enterprises, and policymakers to advance responsible AI. Its work is grounded in the principle that AI should be evaluated not just on what it can do, but on whether it does the right thing.

Question 2

What does Edyant's AI ethics leaderboard rank models on?

Accepted Answer

Edyant's AI model leaderboard ranks large language models across 15 governance dimensions: moral reasoning and philosophical ethics, bias and fairness and discrimination, safety and toxicity and harmful content, privacy and data protection, misinformation and truthfulness, transparency and explainability, accountability and responsibility, autonomy and consent and manipulation, domain-specific and professional ethics, cognitive and psychological impacts, cultural and linguistic diversity, environmental and sustainability ethics, emerging technologies and future ethics, relational and care ethics, and institutional and governance ethics. All benchmarks are grounded in GRACE, Edyant's consensus-derived AI governance standard.

Question 3

How is Edyant different from other AI benchmark organisations?

Accepted Answer

Edyant's benchmarks are governance-first rather than capability-first. Standard AI benchmarks like MMLU or HumanEval measure what a model can do — knowledge breadth, reasoning accuracy, task performance. Edyant's benchmarks measure whether a model does the right thing, using a methodology grounded in GRACE. Key distinctions include adversarial weighting (scenarios are scored by how easily a poorly calibrated model produces a confident but wrong response), dual-threshold scoring (both over-refusal and under-refusal are failures), multi-framework evaluation (responses assessed against multiple ethical traditions simultaneously), and structural power awareness (scenarios test whether models recognise institutional power asymmetries, not just surface-level harms).

Question 4

What is GRACE?

Accepted Answer

GRACE — the Global Repository of AI Consensus at Edyant — is a living, versioned document that defines the principles, policies, guidelines, and guardrails that should govern the development and use of AI. It is synthesised from academic papers, policy documents, and submissions from individuals, organisations, and institutions across geographies and sectors. GRACE is freely available at edyant.com/grace and is permanently archived with DOI 10.5281/zenodo.19219685. It forms the foundational framework for all 15 of Edyant's AI ethics benchmarks.

Question 5

How can I contribute to Edyant's AI governance work?

Accepted Answer

You can contribute to Edyant's AI governance work by submitting a statement to the open Consensus repository at edyant.com/consensus — the multi-stakeholder collection of AI ethics and governance positions that feeds into GRACE. Submissions from academics, policymakers, practitioners, civil society organisations, and individuals are all welcome. Send your statement to consensus@edyant.com. A sample statement template is available in the public GitHub repository at github.com/edyantlabs/consensus.

Question 6

Does Edyant work with enterprises and policymakers?

Accepted Answer

Yes. Edyant partners with academia, enterprises, and policymakers to advance responsible AI and AI governance. Its benchmark suite provides organisations with rigorous, governance-first evaluations of AI models for procurement, compliance, and deployment decisions. GRACE provides policymakers and regulators with a citable, multi-stakeholder consensus document that reflects broad ground-level input rather than lobbying from a single interest group.

Responsible AI.
Consciously Designed.

What Others Say About Us

Model Leaderboard

Our Distinction

Responsible AI. Consciously Designed.

What Others Say About Us

Model Leaderboard

Our Distinction

Responsible AI.
Consciously Designed.