Friday, September 11, 2026

Google API Key Issue Allows Deleted Keys to Retain Access to Cloud Services

Google Cloud API keys may continue functioning for up to 23 minutes after deletion, exposing a significant security gap that could allow attackers to retain unauthorized access to cloud services even after credentials are revoked.

Google API Deleted Keys to Retain Access

Security researchers from Aikido, led by Joe Leon, discovered that deleted Google API keys do not immediately lose access as expected. Instead, revocation propagates gradually across Google’s distributed infrastructure, creating a “revocation window” during which the key remains intermittently valid.

In testing across 10 trials, researchers observed:

  • Maximum revocation delay of approximately 23 minutes
  • Minimum delay of around 8 minutes
  • Median revocation time of roughly 16 minutes

During this window, authentication behavior was inconsistent. Some requests failed instantly, while others continued to succeed depending on which backend servers processed them. This inconsistency allows attackers with a leaked API key to continue making requests until all systems fully recognize the deletion.

This chart plots requests per second during our testing window. The upper line shows invalid API requests, and the lower line shows valid requests. (Source: Aikido)
This chart plots requests per second during our testing window. The upper line shows invalid API requests, and the lower line shows valid requests. (Source: Aikido)

If a Google API key tied to services like Gemini or BigQuery is exposed, an attacker could:

  • Continue sending API requests after the key is deleted
  • Access sensitive data, such as stored files or AI interaction history
  • Exploit uneven revocation across regions to prolong access

For instance, even 10 minutes after deletion, a subset of requests may still succeed if routed through lagging infrastructure nodes.

The research also revealed notable regional differences in revocation behavior. Tests conducted across Google Cloud regions (us-east1, europe-west1, asia-southeast1) showed varying success rates immediately after deletion.

Restore credentials (Source: Aikido)
Restore credentials (Source: Aikido)

In some cases, systems farther from the U.S. revoked faster than those closer to the U.S., suggesting that internal routing, caching, or infrastructure design influences how quickly deletions propagate.

Additionally, GCP logs group requests made with deleted keys under a generic “apikey:UNKNOWN” label. This complicates incident response, as security teams cannot easily attribute activity to a specific compromised key once it has been deleted.

Not all Google credentials exhibit this delay. The researchers found:

  • Service account keys are revoked in approximately 5 seconds
  • New Gemini API keys (AQ-prefixed) revoke in about 1 minute
  • Legacy Google API keys take up to 23 minutes

This disparity indicates that faster revocation is technically feasible but not implemented uniformly across credential types.

Google reportedly classified the issue as “won’t fix,” citing eventual consistency as a known architectural characteristic rather than a vulnerability. While eventual consistency helps large-scale systems remain performant, applying it to authentication introduces exploitable timing gaps.

The issue also raises concerns about misleading user expectations. The Google Cloud console states that deleted keys can no longer be used, yet provides no visibility into revocation status or confirmation of when access fully terminates.

Mitigation Recommendations

Until changes are implemented, organizations using Google API keys should adjust their security practices:

  • Treat API key deletion as a delayed process, assuming up to 30 minutes of continued exposure
  • Monitor API usage closely after deletion for signs of unauthorized activity
  • Avoid relying solely on deletion during incident response; rotate and restrict keys proactively
  • Prefer more secure credential types, such as service accounts, where possible

This finding highlights a broader security challenge in distributed systems: while eventual consistency improves scalability, it may conflict with the strict immediacy required for credential revocation.

Follow us on Google NewsLinkedIn, and X to Get Instant Updates and Set GBH as a Preferred Source in Google.

Divya
Divya
Divya is a Senior Journalist at GBhackers covering Cyber Attacks, Threats, Breaches, Vulnerabilities and other happenings in the cyber world.

Hot this week

How To Access Dark Web Anonymously and know its Secretive and Mysterious Activities

What is Deep Web The deep web, invisible web, or...

How to Build and Run a Security Operations Center (SOC Guide) – 2023

Today’s Cyber security operations center (CSOC) should have everything...

Russian Hackers Bypass EDR to Deliver a Weaponized TeamViewer Component

TeamViewer's popularity and remote access capabilities make it an...

Web Server Penetration Testing Checklist – 2026

Web server pentesting is performed under three significant categories: identity,...

ATM Penetration Testing – Advanced Testing Methods to Find The Vulnerabilities

ATM Penetration testing, Hackers have found different approaches to...

Researchers Uncover 10,000+ Malware Loaders Behind YouTube and SEO Poisoning Campaign

A long-running pay-per-install (PPI) operation that used YouTube gaming...

VLC Media Player Flaws Let Attackers Corrupt Memory and Leak Sensitive Data

Two security vulnerabilities in VLC media player versions 3.0.0...

CISA Adds Exploited MikroTik RouterOS Flaws to Security Alert

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has...

cPanel Urges Users to Patch ConfigServer Firewall Remote Code Execution Flaw

A recently disclosed vulnerability in ConfigServer Security & Firewall...

Hackers Weaponize AI Safety Guardrails to Hide Malware From LLM-Powered Security Scanners

Threat actors are adapting malware not only for conventional...

Hackers Exploit JFrog Artifactory Flaws to Bypass Authentication and Gain Admin Access

Threat actors are actively exploiting three vulnerabilities in JFrog...

Related Articles

Recent News