00 · Information gaps and simulated scores
The simulated risk score covers every scenario category, reflecting missing information and self-reported conflicts. The policy check separately applies current or new-policy scope; upcoming provisions are not treated as already effective.
Per category: undeclared / unknown adds 1 suspicion point; self-reported conflict adds 1; a clear self-declaration adds 0. Count once per category. Simulated score = round(100 × suspicion points ÷ 9). Blank is 100; all items cleared without conflicting declarations is 0.
At 60, the page displays “Simulated review.” This is not an official ban threshold or a claim that an account will be banned immediately. Unknown location or history is treated as an information gap, not a verified official detection rule or evidence of an intermediary account.
The simulation updates live as choices change; the policy check updates on submission. Only structured self-declarations clear the corresponding simulation items. Free text, IP lookup, payment brands, language and nationality do not automatically change suspicion. No legal name, identity documents or account credentials are required. Declarations are self-reports, not evidence or official verification.
01 · How the index is calculated
Identified policy-conflict index = round(100 × flagged categories ÷ total categories evaluated in this mode). Each category counts once. No starting points, billing mismatch points or severity weights.
Current mode has 6 categories; new-policy mode has 9. New region, organization and platform-abuse criteria enter the default assessment on their effective date; earlier previews are clearly labeled.
In the policy check, unknowns add no conflict points but reduce completeness; all unknown means no policy index. The simulation counts unknowns as suspicion instead. The two metrics must not be conflated. With partial answers, the index reflects identified items only; more conflicts may remain unrecognized. Unclassified free text requires manual review and is not keyword-scored.
Each assessment starts fresh; clicks do not accumulate. Explicit conflicts override the numeric index: even one triggers “Explicit policy conflict identified.” Zero does not mean safe. Historical violations, present circumstances and plans are self-reported, not verified facts.
| Category | Flagged when | Applicability | Basis and limits |
|---|---|---|---|
| Authentication | Third-party app collects / intermediates subscription session credentials or routes consumer subscription requests | Currently verified | Official guidance restricts third-party collection or intermediation of Claude.ai credentials and routing on users’ behalf through consumer subscriptions. API keys and supported cloud providers are not categorically prohibited. Claude Code legal and compliance ↗ |
| Account sharing | Shared / rented individual account or individual login given to others | Currently verified | Consumer terms prohibit giving others access to an individual account or credentials. Official organizational seats and authorized commercial credentials are distinct. Consumer Terms of Service ↗ |
| Resale and intermediation | Without special authorization, reselling Claude usage through subscriptions / credentials | Currently verified | Claude Code guidance restricts paying for, reselling or intermediating end-user usage. Special agreements require review. A proxy server alone does not establish a violation. Claude Code legal and compliance ↗ |
| Authentication and protective measures | Unauthorized bypass of protections, authentication or access restrictions | Currently verified | Consumer terms prohibit bypassing systems or protective measures. Claude Code guidance requires preserving the published binary and authentication methods. Normal configuration, MCP and official automation alone do not trigger this item. Consumer Terms of Service ↗ |
| Purpose and content | Explicit illegal activity, fraud, unauthorized intrusion or infringement | Currently verified | Consumer terms prohibit illegal activity, unauthorized access, deception and infringement. Context and authorization matter, not keywords or language. Commercial users must also check applicable commercial terms and usage policy. Consumer Terms of Service ↗ |
| Violation history and continuation | Confirmed repeated violations or continuing the same violations after warnings | Currently verified | Official guidance lists repeated usage-policy violations as a ban reason; counts, scores and decay thresholds are not public. A single refusal, exhausted quota or suspicious activity from compromise does not automatically establish a violation. Safeguards warnings and appeals ↗ |
| Physical-location eligibility | Checked the official list: physical location is unsupported | 2026-11-12 onward; preview before then | The new policy considers physical location. IP does not prove location. Unlisted regions are unsupported. This previews user-verified eligibility; nationality and language are not scored. Supported Regions Policy (new version) ↗ |
| Organizational eligibility | Organization conflicts with incorporation / headquarters / majority ownership or control criteria | 2026-11-12 onward; preview before then | The new policy covers incorporation, headquarters and direct or indirect majority ownership or control. Personal or employee nationality alone cannot establish organizational eligibility. Supported Regions Policy (new version) ↗ |
| Platform abuse (new policy) | Ban evasion, stolen payment, unauthorized distillation or malicious multi-account detection evasion | 2026-11-12 onward; preview before then | The new policy explicitly addresses ban evasion, stolen or fraudulent payment, unauthorized model distillation and malicious multi-account detection evasion. Not all training, batching or multiple accounts are automatically violations; authorization and purpose matter. Usage Policy (new version) ↗ |
02 · Checkboxes, text and risk status
Selects and supplementary checkboxes map to the same categories; duplicates count once. A checked explicit-conflict declaration takes precedence. Contradictory clear selections are flagged. Leaving a checkbox unchecked is not a declaration that no problem exists.
Results are: explicit policy conflict identified, insufficient information / review needed, or no explicit conflict reported. These are checklist outcomes, not percentage-based ban likelihoods. VPN use, travel, authorized automation, payment failures and model refusals are context only.
Example: credential misuse plus individual account sharing gives 2/6 → 33 in current mode, or 2/9 → 22 in new-policy mode. The scope differs; this does not indicate reduced real-world risk.
There are no required fields or minimum text length. Skipped items retain simulation concerns and remain unknown in the policy check. Authentication, history, region, extra text, payment and IP remain optional expandable sections. Text is not automatically scored.
Sources and evidence scope
Claude Code legal and compliance ↗
Public documentation
Consumer Terms of Service ↗
Effective 2025-10-08; consumer accounts
Safeguards warnings and appeals ↗
Help article
Supported Regions Policy (new version) ↗
Effective 2026-11-12; preview before then
Usage Policy (new version) ↗
Effective 2026-11-12; preview before then
Card declines and payment eligibility ↗
Payment processing guidance; not a ban score
Official programmatic usage ↗
Automation support does not authorize every use or credential flow
Usage limits ↗
Quota limits are distinct from account enforcement
Policy conditions come from official sources; equal-category counting and the index formula are this site’s presentation method, uncalibrated against ban data. Anecdotes, VPN use, payment brands, language, nationality and keywords add no points. Current mode is not a reconstruction of the full prior policy: historical region and organization requirements need manual review.
03 · Privacy
Calculated locally in the browser; no Claude connection or saved input. Free text is not sent to a model and is not semantically classified. Optional IP lookup contacts ipapi.co, which receives your public IP; GPS is not used. Hosting may generate ordinary access logs. Do not enter passwords, tokens, card numbers or complete warning emails.
← Back to simulator