Responsible AI Policy

1. How AI Is Used in Scans

AIClicks uses deterministic, rule-based probes for the vast majority of findings: TLS handshakes, DNS queries, header inspection, and CVE correlation. Machine learning is used narrowly and transparently for: (a) AI-content signal analysis across multiple independent methods; (b) prioritizing remediation guidance; (c) summarizing scan evidence in reports. We avoid absolute AI-authorship claims — AI-content results are probabilistic signals, never verdicts.

2. Confidence, Evidence, and Limitations

Every AI-assisted output must include: the confidence level or signal strength, the underlying evidence (which probes and methods produced it), known limitations and false-positive modes, and the context needed to appeal or request human review. Reports state clearly that detection can be wrong. AIClicks reports are evidence-based decision support — not legal proof of AI authorship, fraud, compliance, or security posture.

3. No Autonomous Actions

AIClicks never takes autonomous action on your behalf. Scans only read publicly accessible data (plus verified-domain active tests you explicitly authorize). We do not change your website, contact third parties, file reports, or modify DNS, certificates, or configurations. Remediation is always your decision; our fix recommendations are guidance, not instructions executed by us.

4. Human Oversight

High-impact classifications (for example, AI-content authorship signals used in trust scoring) are designed for human review before any consequential decision. Our knowledge base of vulnerabilities and fixes is curated and versioned by humans. If you believe an AI-assisted result is wrong, contact [email protected] for review — we investigate and correct where warranted.

5. Data Handling

Scan inputs and results are processed under our Privacy Policy and, where applicable, our Data Processing Addendum. We do not train models on your private scan data or sell it. Aggregated, anonymized statistics may be used for research (such as the public security benchmark) only when they cannot identify you or your domains.

6. Transparency Commitments

• No black-box scores: Every finding shows the raw probe output behind it.
• Method disclosure: AI-content analysis names the independent methods used.
• Review path: Every AI-assisted conclusion can be appealed to a human reviewer.
• EU AI Act alignment: We design AI features with the transparency obligations of the EU AI Act in mind, including clear disclosure when content or analysis is AI-assisted.

7. Contact

Questions about how AI is used in AIClicks, or to request review of an AI-assisted result: [email protected]