ChatGPT or Claude Unavailable in Hong Kong? Check These 4 Causes | Free Checklist
“AI isn’t working” can mean four different things: regional eligibility, plan or workspace permissions, usage caps, or an outage. Start by recording the exact error, the time and the product surface you used: web app, mobile app or API.
1. Check regional and business eligibility
As of 12 November 2026, Hong Kong is absent from OpenAI’s published ChatGPT web and mobile list. OpenAI’s separate API list also omits Hong Kong. Anthropic’s policy effective that day covers Claude apps, Claude Code, its API and authorised resellers. Hong Kong is absent from that list. The policy also considers the user’s physical location, the entity’s incorporation and headquarters, and majority ownership or control. Google lists Hong Kong for the Gemini web app. The Google AI Studio and Gemini API list omits Hong Kong. Check mobile and Workspace feature eligibility separately. ChatGPT supported countries OpenAI API supported countries Anthropic Supported Regions Policy Gemini web availability Google AI Studio and Gemini API availability
2. The app opens, but a feature is unavailable
Chat, file uploads, Drive connections and API access can have different eligibility rules. Record the plan and workspace, then ask the administrator to check the setting for the missing feature. Use a dummy file to test opening the feature, reading the file and saving the result as distinct steps. Read access alone gives you no evidence of write access.
3. It worked this morning, then hit a cap
If the product reports a usage limit, record any reset time it shows. Claude usage can depend on conversation length, attachments, the model and tool use. For an API 429, check the error detail and check the rate, quota or spending limit and review billing if the error indicates a billing issue. Save a handoff for work in progress. Claude usage limits Claude API errors OpenAI API errors
4. Several eligible teammates see the same error
Check the provider’s status page for the affected product, time and feature. A problem affecting one teammate warrants checking their workspace, permissions and company network as well. For work close to its deadline, activate the approved fallback and assign someone to record when the team switched and when it recovered. OpenAI status Claude status
Free AI access troubleshooting checklist
Copy these fields into your incident record: date and timezone | product and surface | exact error or code | source for regional and entity eligibility | plan, workspace and administrator check | reset time or status incident | affected work and deadline | next step and owner.
Example: 10:30 HKT | Gemini API | project could not be enabled | check the official API region list separately | administrator check pending | no usage cap shown | ad report due 15:00 | project lead starts the manual reporting process. Redact email addresses, account IDs and keys from any screenshot.
Example: Gemini opens on the web, but the API won’t activate
Check web and API regional eligibility separately. Then review the relevant Cloud project’s plan, billing and administrator settings. Write down the product and evidence for each check so a colleague can continue the investigation without repeating it.
Can we solve this by switching models? Confirm the fallback is available to your company for this use and permitted to handle the relevant data. Run a dummy task through file access, required fields and human review.
Choosing a fallback for your team? Tell Allua which AI capabilities matter, where the team works and what data the workflow handles.




Comments