About IP for AI: Method and Data Limits

Each layer answers a different question

• Network profile: what public network profile best describes this IP: carrier, data center, proxy, Tor, or another type?
• Access profile: when per-IP evidence exists, can the connection be distinguished as fixed broadband, mobile, enterprise, or data-center access?
• Routing profile: do address families or data sources show anycast, cross-region, or region-mismatch signals?
• Risk profile: are public proxy, Tor, or abuse-list signals available and matched?
• AI environment score (0-100): uses shared weights for ChatGPT, Claude, Gemini and Grok to summarize available exit identity, geolocation agreement, reputation and environment consistency. Regional eligibility, DNS and language are assessed separately and are not scored.

Official service status and dual-track diagnostics

We display available official status feeds and sync times. Missing, failed or stale updates remain unconfirmed. A healthy official update does not rule out browser sessions, account eligibility or localized incidents. Historical timelines summarize fetched official incidents, without claiming complete coverage or calculating official uptime.

DNS exit comparison verifies that recursive resolver exits align regionally with browser HTTP exits to prevent cross-border drift. WebRTC candidate discovery detects browser-level public IP exposure. These diagnostic signals explain transport differences and do not decide account access.

How sources are separated

GeoIP sources are used to cross-check region and location. ASN, PeeringDB, and cloud IP ranges support network profiling. Proxy references, Tor exit lists, and public abuse lists provide risk signals. Sources do not override one another: a risk-list match alone does not reclassify an IP as a data-center address.

When sources disagree, the result preserves a review or pending boundary instead of turning uncertainty into an account verdict.

How to read a result

• Confirm the automatically detected IPv4 and IPv6 exits first, then check whether they use different network paths.
• Read network type, routing, risk signals, and the AI environment score together rather than relying on one label.
• AI availability is a browser connectivity result for public resources. For a manually entered IP, it is mainly a regional reference, not a live connection test.
• For account, subscription, organization-policy, or platform-risk issues, retain the original service message and verify it with the relevant provider.

ipforai Network Check

Command-line entry: https://ipforai.cc/sh. Two equivalent runners: curl -fsSL https://ipforai.cc/sh | sh or curl -fsSL https://ipforai.cc/sh | bash. Prefer review first: curl -fsSL https://ipforai.cc/sh -o /tmp/ipforai-check.sh, open with less, then run it with sh or bash.

Developer JSON: GET https://ipforai.cc/json profiles the request source IP observed by the server; GET https://ipforai.cc/json?ip=8.8.8.8 looks up a specified IPv4, and GET https://ipforai.cc/json?ip=2001:4860:4860::8888 looks up a specified IPv6. Copyable examples: curl -sS 'https://ipforai.cc/json', curl -4 -sS 'https://ipforai.cc/json', curl -6 -sS 'https://ipforai.cc/json', and curl -sS 'https://ipforai.cc/json?ip=8.8.8.8'. The only optional parameter is ip; one request normally fills one address family and does not read local system or browser data.

ipforai Network Check is read-only: it does not write local config, install software, or require an API key. It does request this site (and the IPv4 probe host) for public IP profile and probe data. Exit profile and AI environment score use the same backend APIs as the website; the script does not recompute score rules locally. Connectivity and routing probe endpoints are maintained in the script and may differ from browser checks. Results describe the network only, not account, subscription, or platform-risk outcomes.

Public mirror of the CLI docs and script: https://github.com/xjokerabab/ipforai. Runtime and data still use https://ipforai.cc.

Privacy and result boundaries

Results describe this network environment only. The check does not read passwords, verification codes, cookies, subscription state, or internal platform risk data. The diagnostic summary hides the full IP by default; review it before sharing.

No single network signal proves that an account will be banned, a feature will be available, or a platform will reject an IP.

Update policy

Rules, lists, and official regional materials change over time. A page update date describes the update to this site's content or data version; it does not mean a third-party platform made the same decision at that time. When sources conflict or become stale, use the listed source and the latest review.