Content Safety and Online Safety

Our approach

This page describes the measures Lucidpic uses to support compliance with applicable duties under the UK Online Safety Act 2023. We review these measures against relevant Ofcom guidance and Codes of Practice.

Our boundary

Lucidpic is a private-by-default AI image and video creation service. Accounts are available only to people aged 18 or over. We do not permit pornography, explicit nudity, sexual acts, nudification, sexualised intimate imagery, or any sexual content involving a person who is or appears to be under 18. Lucidpic does not provide real-person intimate-image generation. Ordinary real-person image editing is permitted only for lawful, non-explicit purposes. Separately, non-explicit creative content involving entirely fictional adults is permitted. Safety controls prevent real-person images, likenesses, and trained identities from being combined with intimate-image requests.

We prohibit non-consensual intimate imagery, including AI-generated or altered depictions of identifiable people. Users uploading reference images must be authorised to use the image and have the necessary consent to use each person's likeness.

Systems and processes

• Prompts are checked before and after enhancement.
• Reference images sent for generation are checked for explicit content and apparent age.
• Generated outputs pass an always-on content-safety check.
• Images based on reference photos, trained characters, or other real-person likenesses receive additional checks for intimate content, including clothing, pose, and context.
• Media does not appear in our public Explore gallery without a separate safety review, which uses automated safety analysis and human moderation as appropriate.
• Reports and repeated or serious signals can lead to warnings, restrictions, account removal, preservation, or reporting where required.
• Suspected child sexual exploitation and abuse content is restricted and handled under a specialist escalation process. Where reporting duties apply, detected and previously unreported content is reported to the UK National Crime Agency.

We review flagged results and test changes against permitted and prohibited examples to reduce both missed detections and incorrect blocks. Automated checks can make mistakes. Ordinary clothing, conventional swimwear, or an indoor setting alone is not a reason to block an otherwise non-sexual image.

Project Arachnid screening

We use Shield by Project Arachnid, operated by the Canadian Centre for Child Protection Inc., to screen uploaded and reused media for known child sexual abuse material (CSAM) and harmful-abusive material of children. These checks can apply to private media as well as reference images and character or training photos.

A match blocks the upload or processing request. Requests requiring screening also stop if the check cannot be completed or its result cannot be verified. A result with no known match does not establish that content is safe or lawful: known-content matching does not identify all previously unseen abuse material and supplements our other safety checks. Contact us using the review process below if you believe a block was incorrect; do not email suspected CSAM.

Screening sends media to C3P in Canada. Our Privacy Policy explains the information shared and retained.

Image safety checks for developers and businesses

Bring image screening into your own service with the Lucidpic Image Safety API. Built on screening used within Lucidpic, it combines explicit-content screening, a separate content review, and apparent-age assessment. Check images from your own product or workflow and receive structured results to support your team's moderation decisions.

Image assessment is separate from the additional safeguards used when generating content in Lucidpic. A clear result does not verify age or consent, authorize generation, or guarantee legal compliance. Use these signals alongside your own policies and human review where needed.

Explore the Image Safety API, pricing, and integration guide.

Governance and review

We maintain documented illegal-content risk assessments and records of our safety measures. We review them at least annually and after significant service changes, serious incidents, or material changes to applicable guidance.

Evidence and privacy

We keep a restricted incident register so authorised staff can monitor controls, review repeated signals, record decisions, and assess whether safeguards remain effective. The register records operational facts such as the control used, reason code, time, severity, linked account or generation, and any review action.

The incident register does not duplicate prompt text, image files, raw IP addresses, or provider payloads. Network data used for abuse correlation is one-way pseudonymised and access is restricted.

Retention

• Files confirmed as prohibited by our output safety check are permanently deleted no later than 30 days after classification.
• One-way network fingerprints in the incident register are removed after 90 days.
• Privacy-minimised incident records are normally retained for up to 24 months to support abuse prevention, control review, and regulatory accountability.
• Relevant information may be retained longer under a documented legal hold, credible illegal-content report, or binding legal requirement.

Reporting and corrections

Report illegal or prohibited content to hello@lucidpic.com. If a lawful fashion or editorial image was incorrectly blocked, contact us so we can review the decision and improve the distinction between prohibited and permitted content.

Reports involving suspected child sexual exploitation and abuse or non-consensual intimate imagery are prioritised. A depicted person or authorised representative can use our public intimate-image reporting form without an account. We remove or disable qualifying reported content and known identical copies within 48 hours, and add image matches to a prevention register checked against future uploads and generated outputs. Email remains available as a fallback.

Explore accepts only content that passes our public-gallery safety standard. If your submission is rejected, or your post, comment, or account is actioned on content or safety grounds, you may appeal by emailing hello@lucidpic.com within 30 days of the action.