BUG: "Failed to generate content. Permission denied" Across All Models

Gemini API
1

Google AI Studio web interface suddenly stopped working for my account after previously working normally. As of April 18, 2026, AI Studio was working fine. My last successful work in AI Studio involved audio description scripts for accessibility-related writing. The content was not sexual, violent, abusive, spam, credential-harvesting, or otherwise policy-violating.

After that, AI Studio began failing. The issue is not limited to one prompt, one model, one browser, one API key, or one Cloud project. AI Studio now blocks generation and even blocks creating a new AI Studio project.

Observed errors:

  1. In AI Studio Playground, generation fails with:
    “An internal error has occurred”
    and/or
    “failed to generate content. permission denied.”

  2. Chrome DevTools → Network shows the GenerateContent XHR request returning HTTP 403.

  3. The GenerateContent response body includes:
    [,7,“The caller does not have permission”]

  4. CountTokens succeeds with HTTP 200, but GenerateContent fails with HTTP 403.

  5. Creating an API key directly inside AI Studio fails with:
    “failed to generate API key. The request is suspicious.”

  6. After deleting all API keys, removing both projects from AI Studio, and trying to create a new project directly inside AI Studio, project creation also fails with:
    “Unable to create project. The request is suspicious.”

  7. Chrome DevTools → Network for the failed AI Studio project creation request shows:

[
7,
“Suspicious activity detected. Request blocked.”,
[
[
“type.googleapis.com/google.internal.alkali.applications.makersuite.v1.AiStudioErrorDetails”,
[
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
[
1
]
]
]
]
]

What I tested:

  • Tried Chrome normally.

  • Tried Chrome incognito.

  • Tried Microsoft Edge.

  • Enabled DevTools Preserve Log and Disable Cache.

  • Tested with a harmless prompt: “This is a test.”

  • Tested with Gemini 3 Pro Image / Banana Pro.

  • Tested with a normal text model: gemini-3.1-flash-lite.

  • Turned off tools and grounding:

    • Google Search grounding off

    • Google Maps grounding off

    • URL context off

    • Function calling off

  • Same GenerateContent 403 occurs.

  • Checked the Response tab in Chrome DevTools.

  • Confirmed CountTokens returns 200 while GenerateContent returns 403.

  • Confirmed AI Studio History/ListPrompts still loads.

  • Confirmed AI Studio can show older history items, including old work from weeks/months ago.

  • Tried creating a new API key from AI Studio; blocked as suspicious.

  • Tried creating API keys from Google Cloud Console; Cloud Console key creation worked.

  • Created a new Google Cloud project.

  • Added/checked IAM roles, including API-related and Service Usage-related roles.

  • Created a service account and bound new API key to it per latest changes.

  • Confirmed billing is postpay and has a credit balance available.

  • Deleted all API keys.

  • Removed both projects from AI Studio.

    • Tried creating a fresh project directly in AI Studio; blocked as suspicious.

    • Tried creating a fresh project from cloud console, and that works fine.

    • This is not happening only with one old project or one old API key.

A Cloud Console-created Gemini API key worked outside AI Studio. I successfully called the Gemini REST API from PowerShell using generateContent. The REST API returned candidates and usageMetadata.

This suggests:

  • My Google account is not globally disabled.

  • Google Cloud Console still works.

  • Cloud project/API key creation through Cloud Console works.

  • The Gemini API runtime can work outside AI Studio.

  • Billing is not globally blocking all Gemini API usage.

  • The problem is specific to AI Studio’s web/control-plane flow.

Current behavior:

AI Studio can load some read/history operations, such as History/ListPrompts, but blocks GenerateContent and blocks project creation. The backend response explicitly says:
“Suspicious activity detected. Request blocked.”

Expected behavior:

AI Studio should allow me to create/open a project and generate content, or it should show a clear, actionable account/project remediation path. If an automated security system has falsely flagged my account/project/session, there should be a review or appeal path.

Why I believe this is a false positive or backend entitlement/provisioning issue:

  • The last successful AI Studio content was accessibility-related audio description script work.

  • The issue now happens even with the harmless prompt “This is a test.”

  • It happens before prompt content is relevant, because creating a new AI Studio project is also blocked.

  • It happens across Chrome, Chrome incognito, and Edge.

  • It happens with tools/grounding disabled.

  • It happens with both image and normal text models.

  • Cloud Console and direct Gemini API access are not globally blocked.

  • The DevTools response explicitly indicates an AI Studio/MakerSuite internal suspicious-activity block:
    type.googleapis.com/google.internal.alkali.applications.makersuite.v1.AiStudioErrorDetails

This does not appear to be a prompt safety issue, a browser issue, an API key issue, a billing issue, or a normal Cloud IAM issue. It appears to be an AI Studio-specific web interface backend authorization/provisioning bug.
Does anyone know either how to fix it/work around it , or alternatively, who I should call or email at Google to fix it?
I’m honestly about to write Google off. This is absurd and beyond the level of stupidity I should be expected to handle. Google’s lack of communication and transparency with respect to contact methods for fixing these sorts of things is astounding to me. Consumers really shouldn’t be forced to put up with this level of ineptitude.