Knowledge base · TOPIC-16

Captive portal or AI provider failure?

Short answer: establish general Internet access before attributing an AI failure. A portal is a network-admission condition; a provider incident is a separate hypothesis.

A useful order is: physical link, local address, gateway, recent network transition, system captive indication, approved plain-HTTP connectivity behavior, then provider checks. Redirect or interstitial evidence raises portal confidence. Repeated Internet failure plus a new Wi-Fi path can justify attention, but it should not automatically disable security controls.

Complete portal authentication manually. Never maintain a universal portal-domain keyword list, submit credentials automatically, or accept terms on the user’s behalf. Once Internet access is restored, record provider evidence and official status independently. If only one provider fails, a portal is less likely, but root cause remains unproven without more data.

Open Probe V0.1 does not detect or bypass captive portals automatically. It is a read-only local evidence tool. Use the Scanner before and after authentication, and Open Probe when local transport context is needed. Related: public Wi-Fi and AI tools.