Safety is built into how we develop and design our AI products, not added on as an afterthought. We use a safety-by-design approach at all stages of product development and deployment to ensure that our platforms are used as intended.
1. Zero-tolerance policy for child safety violations
Any attempt to generate content that sexualizes, exploits or resembles minors is strictly prohibited. Our moderation efforts are designed to detect and prevent such attempts. We proactively detect attempts to create such content; and when CSAM is identified, we:
- Report it to the National Center for Missing and Exploited Children (NCMEC);
- Permanently ban the accounts involved;
- Cooperate fully with law enforcement investigations.
This is an absolute rule. There are no exceptions.
2. Robust technical & organizational content moderation measures
We proactively and reactively moderate content, using a combination of technology tools and human review.
We invest considerable resources into developing moderation technologies. These tools are continuously refined by our in-house teams based on insights gained from observed platform usage and our participation in the trust & safety community, for example, the keywords list that Internet Watch Foundation makes available to its members. Additionally, users can report content they find concerning via in-product report buttons.
Decisions to terminate account access based on policy violations are always made by human reviewers. Human reviewers may take a variety of actions against violative content, including:
- Temporary or permanent bans
- In appropriate cases, referral to law enforcement or NCMEC.
3. Full transparency about our policies
Our Community Guidelines, Blocked Content Policy, Underage Policy and all related legal documents are publicly available on our websites
We update these policies as our platform evolves and as the legal and safety landscape changes, and we publish annual transparency reports.
4. User responsibility and accountability
Users are ultimately responsible for the interactions and content they initiate. Attempts to misuse the platform or circumvent our safety systems are handled in line with our moderation policies.
5. Mental health and wellbeing
We design AI companions to support positive user experiences. Our products and services are not a substitute for professional mental health care. If you are feeling distressed or you are experiencing mental health difficulties, please reach out to a qualified professional for assistance.
Via our partnership with Throughline, we have implemented technical measures to offer localized crisis resources where users are identified as expressing thoughts of self-harm.
6. Ongoing evaluation
Our safety technologies and policies are reviewed on an ongoing basis and kept up-to-date based on platform changes, regulatory developments, and findings from our moderation systems.
7. External engagement
We are active in and welcome dialogue with the trust & safety community. See more information on our partnerships here.
Comments
0 comments
Article is closed for comments.