AI Transparency Statement

• You upload or photograph a chemical label from within the Add or Edit chemical dialog.
• The image is resized client-side to reduce data transfer.
• The image is sent to the Anthropic Claude API via LevelSixLabs's servers — it does not go directly from your browser to Anthropic.
• Claude analyses the image and returns structured field suggestions.
• LevelSixLabs presents these suggestions in a checklist — nothing is applied to the form automatically.
• You review each suggestion, tick or untick fields, and apply only what you choose.
• Fields with existing values are unticked by default, flagged with a warning so you can decide whether to overwrite.

• AI can misread handwritten, damaged, or low-contrast labels
• Hazard codes and GHS classifications should always be verified against the official SDS
• CAS numbers, expiry dates, and quantities should be confirmed before use
• AI does not have access to regulatory databases — it reads what is visible on the label only
• Confidence is not guaranteed even when the scan appears complete

• AI suggestions are always presented as suggestions — never auto-applied
• Every field is individually selectable before applying
• You can reject all suggestions with a single click
• You can edit any applied value before saving
• AI scan results are not stored — only the fields you choose to apply are saved
• You can use LevelSixLabs entirely without AI features

When you use label scanning, the image is sent to Anthropic, PBC via their Claude API. Anthropic processes the image solely to generate the response and does not use LevelSixLabs customer data to train its models.

Images are not stored by LevelSixLabs or Anthropic after the request is complete. Only the field values you choose to apply are saved to your LevelSixLabs database.

Anthropic is listed on our Subprocessor List.

• We clearly label all AI-generated content in the UI
• We do not use your data to train AI models
• We do not make automated decisions with legal or significant personal effect based solely on AI outputs
• We review AI features periodically to ensure they remain safe and useful
• We will disclose any new AI features in this statement before they are released

LevelSixLabs's AI features are designed with UK GDPR, the EU AI Act, and Health & Safety at Work Act principles in mind. Chemical hazard information generated by AI is presented as a starting point for human review — not as a substitute for professional assessment or official SDS documentation.