← Best Practices
ETHICS AND COMPLIANCE

Ethical Considerations When Using ESM3

Ethical ESM3 research in 2026: dual-use biosafety, data rights, evaluation bias and honest scientific communication.

Updated for 2026: Ethical ESM3 research in 2026: dual-use biosafety, data rights, evaluation bias and honest scientific communication.

Assess dual-use risks

Protein design can be beneficial but may be misused. Apply institutional biosafety procedures and provider acceptable-use policies. Requests to increase the harmful effects of biological agents need expert risk assessment and may be prohibited.

Respect data permissions

Confirm that training and inference inputs may legally be processed by the selected provider. Avoid unnecessary transmission of proprietary or personal information and review platform retention policies.

Avoid overstatement

Model outputs are hypotheses, not validated treatments, environmental solutions or safety guarantees. Disclose computational limitations, known evaluation gaps and uncertainty.

Keep people accountable

Document human review for sensitive studies and make responsibility for decisions explicit. Follow relevant local institutional, export-control and applicable regulatory requirements.

Evidence and further reading

Primary sources: Biohub ESM3 model, Biohub maintained ESM repository, and Biohub learning resources. Check model cards and release notes for version-specific instructions.

Editorial note: Updated October 2026. Distinguish scientific literature and current product documentation from speculative applications; verify current API details before running code.

Scientific context

Research education only. Outputs from protein AI models are predictions and require independent evaluation. Verify APIs, model availability and licensing with official documentation.

✦ AI RESEARCH ASSISTANT

Questions about this article?

Get a concise explanation grounded in ESM3 Academy articles, with links to the sources. No need to leave this page.

Ask AI about this article ↗