AI & Synthetic Media Policy

Zeeker uses machine learning and AI to support recommendations, search, moderation, translation, summaries, assistance, safety, and product development.

Where AI may be used

• rank feeds, topics, communities, and search results;
• detect spam, coordinated manipulation, abuse, or unsafe content;
• assist reviewers with prioritization and context;
• generate summaries, translations, labels, prompts, and responses;
• ground answers in linked or licensed sources;
• improve reliability and evaluate features; and
• train or improve systems using licensed, permitted, de-identified, synthetic, or public user content as described in the User Terms.

AI output can be wrong, incomplete, biased, or outdated. It is not legal, medical, financial, or other professional advice. Important decisions require human judgment.

Public user content and training

The User Terms grant a broad license for public user content, including AI/ML training and evaluation. Nonpublic messages and restricted content are excluded from public-data products and general-purpose training. Personal-data rights remain unaffected.

Publisher material is different: Zeeker uses it for training, RAG, datasets, or licensing only when the source ledger records permission or another applicable legal basis covering that exact use. A publisher can request separate controls under the Publisher Data Policy.

Synthetic-media disclosure

Users must clearly label content that is materially generated or manipulated when it could reasonably be mistaken for authentic evidence, a real person saying or doing something, or a real public event. Prohibited synthetic content includes child sexual abuse material, non-consensual intimate imagery, fraud, harmful impersonation, fabricated evidence used to harass, and undisclosed civic manipulation.

Zeeker may add or preserve a label, provenance signal, watermark, content credential, or disclosure. Removing or falsifying a required provenance indicator may violate the Community Guidelines.

Automated moderation

Automation may identify and prioritize suspected violations, but a signal is not conclusive. Consequential complaints receive human review where required. Affected people can appeal and learn whether automation materially influenced the decision, unless disclosure would compromise security or enable evasion.

Sensitive and prohibited uses

Zeeker data and AI features must not be used for:

• facial recognition or persistent biometric identification;
• surveillance of vulnerable people;
• credit, employment, housing, insurance, criminal-risk, or similar high-impact eligibility decisions;
• inferring sensitive traits for discrimination or unlawful advertising;
• harassment, doxxing, manipulation, or weapons targeting; or
• evading publisher, user, safety, consent, or deletion controls.

Questions and reports

Report misleading synthetic media through the illegal-content form. Publishers can specify AI-answer, training, and licensing controls through the publisher form.