How we protect the network

Trust and safety at Lona rests on three commitments: real experts doing real work, careful handling of sensitive material, and quick response when something goes wrong.

We do not treat trust as a one-time check at sign-up. It is a continuous practice — we verify identity, audit work, run peer review, monitor for abuse, and partner with clients on responsible-use review of every project before it launches.

Every applicant goes through structured verification before they can claim a paid task.

• Identity verification through licensed third-party providers.
• Credential checks for regulated domains (medicine, law, finance, engineering) using authoritative registries.
• Domain assessments to confirm declared expertise.
• Ongoing reverification for high-risk tasks and senior tiers.

We turn down most applicants. That is by design — the value of the network depends on who is in it.

Every project on Lona is reviewed before it is offered to experts. Reviewers consider:

• The purpose of the project and whether it advances responsible AI development.
• The sensitivity of the material the task may surface.
• The skills, tooling, and emotional readiness needed to do the work well.
• Whether the project requires regional, age, or credentialled gating.

Experts can opt in or out of any project, and can decline specific tasks within a project without penalty.

Some projects — particularly red-team work, safety evaluation, and content moderation — require exposure to material that can be difficult to engage with. We take care of the people doing this work.

• Pre-onboarding briefing on what to expect and how to step back.
• Session-time limits and break reminders.
• Optional access to confidential wellbeing support through our partners.
• Anonymous feedback channels so experts can raise concerns about a project.

Client material, expert submissions, and quality reviews are encrypted in transit and at rest. Access inside Lona is restricted to staff with a documented need. We log access to sensitive material so we can audit it.

Experts agree to confidentiality before being shown project briefs. Sharing client material outside the platform is a serious violation of our terms.

If you encounter abuse, harassment, a security issue, or content that should not be on the platform, report it from your account, by email to trust@lonaos.com, or through our contact form.

We acknowledge reports within one business day and aim to resolve material issues within seven business days. We follow regulated disclosure timelines for incidents that affect personal data.

We will publish an annual Trust & Safety report summarising the categories of incidents reported, our response times, and the steps we have taken to improve. The first report will be available at the end of our next fiscal year.