Hackers News Hackers News
  • CyberSecurity News
  • Threats
  • Attacks
  • Vulnerabilities
  • Breaches
  • Comparisons

Social Media

Hackers News Hackers News
  • CyberSecurity News
  • Threats
  • Attacks
  • Vulnerabilities
  • Breaches
  • Comparisons
Search the Site
Popular Searches:
technology Amazon AI
Recent Posts
How Tier 1 Can Process Alerts 3x Faster with Threat Intel
May 26, 2026
Angular Language Service Flaws Enable Remote Code Execution
May 26, 2026
Quasar Linux RAT Targets Developers via Fileless eBPF
May 26, 2026
Home/CyberSecurity News/Deleted Google API Keys Still Access Gemini, BigQuery,
CyberSecurity News

Deleted Google API Keys Still Access Gemini, BigQuery,

Google Cloud API keys, even after deletion, can remain active for up to 23 minutes, a newly disclosed issue reveals. This vulnerability exposes projects to potential abuse long after credentials have...

David kimber
David kimber
May 22, 2026 3 Min Read
22 0

Google Cloud API keys, even after deletion, can remain active for up to 23 minutes, a newly disclosed issue reveals. This vulnerability exposes projects to potential abuse long after credentials have been revoked.

The finding raises concerns about delayed credential invalidation across Google’s infrastructure, particularly for sensitive services such as Gemini, BigQuery, and Google Maps APIs. According to Aikido research, deleting a Google API key does not immediately terminate its access.

Instead, revocation propagates gradually across distributed systems, creating a “revocation window” during which the key continues to authenticate requests.

  • Longest observed window: ~23 minutes.
  • Shortest observed window: ~8 minutes.
  • Median duration: ~16 minutes.

Attackers with leaked keys can continue making API calls during this period because some backend servers may still accept deleted keys, causing inconsistent enforcement.

Deleted Google API Keys Continue Access

The issue becomes more severe when high-value services are enabled. If a compromised key has access to Google’s Gemini API, attackers may:

  • Retrieve previously uploaded files.
  • Access cached conversations.
  • Continue interacting with AI endpoints.

Similar behavior was observed across other services, including the BigQuery and Maps APIs, indicating that the delay is tied to API key infrastructure rather than individual services.

Chart shows invalid requests above valid API requests during testing. (Source: Aikido)
Chart shows invalid requests above valid API requests during testing. (Source: Aikido)

Researchers conducted controlled experiments over multiple days:

  • Created and deleted API keys in repeated trials.
  • Sent 3–5 authenticated requests per second post-deletion.
  • Measured how long requests continued to succeed.

Results showed unpredictable success rates. For example, one minute after deletion, some trials still saw up to 79% of requests succeed, while others dropped to as low as 5%.

This inconsistency makes it difficult to determine when a key is truly invalid. Tests across multiple Google Cloud regions revealed uneven propagation:

  • us-east1: ~49% median success rate.
  • Europe-west1: ~49% median success rate.
  • asia-southeast1: ~22% median success rate.

Interestingly, some distant regions rejected deleted keys faster than closer ones, suggesting that routing, caching, or infrastructure differences influence revocation timing.

The Google Cloud Console does not clearly indicate that a deleted key is still active. Instead:

  • Deleted keys disappear from the interface immediately.
  • Ongoing requests may still succeed without visibility.
  • Failed requests are grouped under “apikey:UNKNOWN”.

This aggregation complicates incident response, as security teams cannot easily attribute activity to a specific deleted key.

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

Not all Google credentials exhibit the same delay:

  • Service account keys: revoked in ~5 seconds.
  • New Gemini API keys (AQ prefix): revoked in ~1 minute.
  • Legacy API keys: up to 23 minutes.

This disparity suggests that faster revocation is technically feasible but not implemented for standard API keys.

Aikido researcher Joe Leon said Google marked the issue as “won’t fix,” describing the delay as expected behavior in eventually consistent systems rather than a security flaw.

While Google documents eventual consistency in IAM systems, it does not explicitly warn users about delayed API key revocation.

Security Implications

Delayed revocation contradicts typical expectations that deleting credentials immediately blocks access. Even short delays can be exploited, as prior cloud security research demonstrates.

For organizations using Google Cloud, this creates several risks:

  • Continued access after credential compromise.
  • Lack of visibility into active misuse.
  • Difficulty enforcing just-in-time (JIT) credential strategies.

Until changes are implemented, security teams should adjust their response strategies:

  • Treat API key deletion as a 30-minute process rather than an immediate one.
  • Monitor API usage closely after deletion for suspicious activity.
  • Rotate keys proactively and minimize exposure in public repositories.
  • Prefer service account keys or newer credential types where possible.

This discovery highlights a broader challenge in cloud security: balancing scalability with strict authentication guarantees. In the case of Google API keys, the current model leaves a critical gap that attackers can exploit during the revocation window.

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.

Tags:

AttackExploitSecurity

Share Article

David kimber

David kimber

David is a penetration tester turned security journalist with expertise in mobile security, IoT vulnerabilities, and exploit development. As an OSCP-certified security professional, David brings hands-on technical experience to his reporting on vulnerabilities and security research. His articles often feature detailed technical analysis of exploits and provide actionable defense recommendations. David maintains an active presence in the security research community and has contributed to multiple open-source security tools.

Previous Post

CISA Adds Langflow Flaw to Exploited Vulner Origin Validation

Next Post

LiteSpeed cPanel 0-Day Exploited for Server Plugin Gain

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Popular Posts
Microsoft Defender Isolates Compromised Devices from Ransom
May 26, 2026
GitHub Down: Authentication Issues Deny Access to Actions
May 26, 2026
Hackers Exploit Ghost CMS CVE-2026 CVE-2026-26980 Poison
May 26, 2026
Top Authors
Marcus Rodriguez
Marcus Rodriguez
Jennifer sherman
Jennifer sherman
Emy Elsamnoudy
Emy Elsamnoudy
Let's Connect
156k
2.25m
285k

Related Posts

Jennifer sherman
By Jennifer sherman
Threats

GlassWorm Attacks macOS via Malicious VS Code…

January 1, 2026
Emy Elsamnoudy
By Emy Elsamnoudy
Attacks

ClickFix Attack Hides Malicious Code via Stegan Security

January 1, 2026
Sarah simpson
By Sarah simpson
Vulnerabilities

MongoBleed Detector Tool Detects Critical MongoDB CVE-

January 1, 2026
Emy Elsamnoudy
By Emy Elsamnoudy
Breaches

Conti Ransomware Gang Leaders & Infrastructure Exposed

January 1, 2026
Hackers News Hackers News
  • [email protected]

Quick Links

  • Contact Us
  • Privacy Policy
  • Terms of service

Categories

Attacks
Breaches
Comparisons
CyberSecurity News
Threats
Vulnerabilities

Let's keep in touch

receive fresh updates and breaking cyber news every day and week!

All Rights Reserved by HackersRadar ©2026

Follow Us