1. Reporting period

Reporting period: 1 January 2026 to 31 December 2026 (first full year of operation; partial-year coverage applies from launch date through 31 December 2026).

This is the public skeleton. Aggregated numbers are populated by an internal job after each reporting period closes. Until the first period closes, fields below carry placeholder values rather than fabricated figures.

2. Orders received from member state authorities (DSA art. 9 and 10)

Orders to act against illegal content (art. 9) and orders to provide information (art. 10), broken down by member state and by category of illegal content.

Median response time: not applicable (no orders received). The internal SLA target is 24 hours for art. 9 orders on harmful content, 1 hour for TCO Regulation 2021/784 terror content removal orders, and 7 days for art. 10 information requests.

3. Content moderation at our own initiative

Actions taken by SynthCamp moderators on our own initiative (i.e., not in response to an order or a notice), broken down by type of action and legal/contractual basis. Internal-initiative actions include catalog audits, abuse-signal triage, and repeat-infringer reviews.

4. Notices and complaints received

Total volume of notices received under DSA art. 16 (illegal content reporting) and conformity reports, broken down by category. Includes the median time from notice to decision.

5. Internal complaint-handling (DSA art. 20)

Number of complaints filed by users against our moderation decisions (appeals) through the internal complaint-handling system at /account/appeals, and median time from filing to resolution.

6. Use of automated means

SynthCamp uses automated tooling at two stages of the moderation pipeline. None of these tools take final decisions; every sanction is confirmed by a human moderator.

• AI provenance scan at upload. Audio uploads are inspected for ID3v2.4 user-defined text frames, watermark signatures, and known-model fingerprints. Results feed the Creative Credits validation surface but never auto-reject a release.
• Audio fingerprint deduplication. Uploaded audio is fingerprinted against the ACRCloud catalog to surface near-duplicates of existing releases for moderator review.
• Anti-coordination signals. When a report lands, the moderator dashboard surfaces aggregate account-overlap indicators to help distinguish brigading from organic reports. The specific signals are operational and are not published so they cannot be used to calibrate evasion; the indicators NEVER pre-classify the decision, they are read by the moderator who decides.
• Burst detection. Clusters of reports filed within a short window against the same target trigger higher-scrutiny review of the reporters. The cluster size and window are operational and are not published. Same caveat: signal for the human, not a decision.

Accuracy of each automated surface (false-positive rate against the moderator's final decision) is tracked internally and is summarized in this report once a full reporting period closes.

7. Additional SynthCamp disclosures

Beyond the DSA minimum, we publish:

• Total volume of reports received, by category (surfaced in section 4 once data lands)
• Sanctions overturned on appeal, as a quality signal on the original decision (surfaced in section 5 once data lands)
• SynthCamp Priority Flaggers count. Our internal program of manually-designated good-faith reporters, distinct from DSA art. 22 trusted flaggers. See Community Standards section 7.6 for the contractual difference.

8. How this report is generated

A scheduled job runs in the first week of each January and aggregates the previous year from the moderation decision log, the notices table, and the internal-complaints table. The job produces the figures above; no manual editing of post-period numbers takes place.

The raw data is auditable internally on request from a competent authority. Personal data identifying reporters or reportees is not exposed in this report. Aggregations preserve k-anonymity by suppressing categories with fewer than five underlying events.

9. Previous reports

Previous yearly reports are archived at /legal/transparency/moderation/history (link active once the first archived report exists).