**Pedalgo FAQ**

**English Version**

**Q1: What is Pedalgo?**  
Pedalgo is a research-oriented triage system that uses AI to surface potential proactive and reactive predatory signals on social platforms. It reorders a human review queue; trained reviewers always make the final decision.

**Q2: What is Pedalgo not?**  
Pedalgo is not an enforcement system, not a public naming tool, and not a vigilante platform. It does not block accounts, issue sanctions, or publish individual flags.

**Q3: Why are most flags expected to be false positives?**  
Because the base rate of actual predatory behavior is low, any detection system produces far more false positives than true positives. Scores are risk signals only.

**Q4: What is the role of humans?**  
Humans conduct all final review and decision-making. AI output serves solely to prioritize the queue.

**Q5: Why does Pedalgo use one universal base rate?**  
A single prevalence rate is applied across all populations. No ethnicity, nationality, or group weighting is used.

**Q6: How does Pedalgo differ from platform moderation?**  
Platform moderation can act directly on accounts. Pedalgo only provides a ranked list for human experts and routes confirmed high-risk cases through lawful channels (NCMEC CyberTipline, Swedish Police, IWF, DSA trusted flagger mechanisms).

**Q7: What happens to data?**  
Data is processed only for research and triage purposes under applicable law. No public disclosure of individual cases occurs.

**Q8: How can someone contest a flag?**  
Contact the relevant platform or authority through their official appeal or notice-and-takedown processes. Pedalgo itself does not maintain public case records or perform direct user appeals.

**Q9: What Sweden-specific resources exist?**  
Swedish users should contact Polisen (Swedish Police) via their online reporting form, the Swedish Media Council, or established hotlines such as BRIS. Lawful escalation also includes the IWF and NCMEC when cross-border elements are present.

**Q10: Are the numbers and examples real?**  
All statistics, prevalence figures, and case examples remain illustrative and synthetic until real platform evaluation data with reviewer-confirmed labels exists.

**Q11: Does Pedalgo publish keyword lists?**  
No. Keyword lists or detection patterns that could be easily circumvented or misused are never published.

**Q12: What is the governance firewall?**  
It is the strict separation between AI triage output and any enforcement action. Only humans decide; AI never triggers automated measures.

**Q13: How are algorithms made available?**  
Research algorithms and methodology documentation are released publicly once they meet transparency and safety review standards.

**Q14: Can Pedalgo be used for vigilante action?**  
No. Any such use violates the project’s terms and legal framework. All outputs are intended exclusively for trained reviewers and lawful authorities.

**Q15: What is the process for team setup guidance?**  
Detailed guidance covers reviewer training, conflict-of-interest rules, escalation protocols, and documentation standards. It is available to verified research and safety organizations.

**Q16: How does Pedalgo support DSA trusted flagger channels?**  
It is designed to feed into existing trusted flagger mechanisms rather than replace them.

**Q17: Are scores determinations of guilt?**  
No. Scores are never presented as proof of any offense; they are prioritization signals only.

**Q18: What happens to synthetic examples?**  
All examples are clearly labeled “SYNTHETIC — FOR ILLUSTRATION ONLY — NOT REAL CASES” and do not represent actual individuals or events.

**Q19: How can researchers access the research hub?**  
Public methodology documents, synthetic datasets, and citation lists are published on the site. Access to raw evaluation data requires formal agreements and ethics approval.

**Q20: Why focus on Sweden as an example jurisdiction?**  
Sweden provides a clear regulatory environment (DSA implementation, national reporting structures) that allows transparent documentation of the full triage-to-escalation pipeline.

**Q21: Does Pedalgo detect both proactive and reactive behavior?**  
The system is designed to surface signals associated with both grooming/initiation patterns and reactive engagement with existing harmful content, always subject to human review.

**Q22: Where should concerns about potential child safety issues be reported?**  
Report directly to the platform, Swedish Police, NCMEC CyberTipline, or IWF. Pedalgo does not accept direct public reports.

---

**Svensk version**

**F1: Vad är Pedalgo?**  
Pedalgo är ett forskningsinriktat triage-system som använder AI för att lyfta potentiella signaler om proaktiv och reaktiv predatoriskt beteende på sociala plattformar. Systemet rangordnar en kö för mänsklig granskning; utbildade granskare fattar alltid det slutliga beslutet.

**F2: Vad är Pedalgo inte?**  
Pedalgo är inte ett verkställighetssystem, inte ett verktyg för offentlig namnpublicering och inte en vigilante-plattform. Det blockerar inte konton eller utfärdar sanktioner.

**F3–F22** följer samma struktur som den engelska versionen ovan, med exakt samma innehåll översatt till korrekt, neutral svenska. Alla svar betonar att AI endast prioriterar, att människor alltid beslutar, att exempel är syntetiska och att all eskalering sker via lagliga kanaler (Polisen, NCMEC, IWF, DSA).