Sources & Methodology
The figures cited on joinhalo.dev come from two places: public research and regulatory records, and our own internal evaluation. This page documents both. Where a number is an estimate or an internal measurement, we say so plainly.
Public figures
| Figure on the site | Source & method |
|---|---|
| Hours children spend on screens each year | Estimated by combining average daily screen-time figures from Common Sense Media's media-use research with the U.S. child population from the U.S. Census Bureau (average daily hours × population × 365). Presented as an order-of-magnitude estimate. |
| New apps and updates shipped each year | Annualized from public app-store release and update volumes reported by Appfigures and Statista. Presented as an order-of-magnitude estimate. |
| FTC fines across children's-privacy enforcement actions | Aggregated civil penalties from U.S. Federal Trade Commission settlements and orders in children's-privacy (COPPA) matters, per the FTC's public children's-privacy enforcement record. |
| 13+ jurisdictions with children's-protection rules | Count of distinct legal frameworks Halo maps against. See the list below. |
Jurisdictions Halo maps against
The "13+" figure reflects distinct children's data-protection and online-safety frameworks, including:
- United States (federal): Children's Online Privacy Protection Act (COPPA) and the FTC's 2025 amendments.
- California: CCPA/CPRA and the California Age-Appropriate Design Code (CA AADC).
- Other U.S. states with youth-privacy or design-code provisions (e.g., Utah, Texas, Connecticut, and others).
- European Union: GDPR (including Article 8 on children's consent) and the EU AI Act's provisions affecting minors.
- United Kingdom: UK GDPR and the Age-Appropriate Design Code (Children's Code).
- Ireland: the Data Protection Commission's Fundamentals for a Child-Oriented Approach to Data Processing.
- Additional national regimes across other regions that impose children's-privacy or online-safety obligations.
Internal metrics
Some figures reflect Halo's own work rather than third-party research. We present these conservatively and can share methodology on request.
| Figure on the site | How it is measured |
|---|---|
| 180+ safeguards | The number of distinct checks in Halo's internal safeguard library, mapped from the statutory and regulatory frameworks above plus child-development and AI-safety research. |
| ~90% accuracy | Result of internal evaluation on validated benchmark slices of real and synthetic codebases. This is an internal measurement on a controlled benchmark, not an independently audited or production figure. Methodology available on request. |
| "Controlled beta" performance claims | Observations from internal, controlled pull-request review tests (flagging real issues, staying quiet on clean changes, and suppressing duplicate noise). These describe current testing, not certified product performance. |
A note on estimates
Where a figure is described as an estimate, it is intended to convey scale, not a precise count. Public datasets are updated over time, and we refresh our figures periodically. If you are evaluating Halo and need underlying numbers or methodology in more detail, email us at contact@joinhalo.dev.