BotBeat
...
← Back

> ▌

Google / AlphabetGoogle / Alphabet
RESEARCHGoogle / Alphabet2026-05-21

Security researchers discover 23-minute vulnerability window in Google API key revocation

Key Takeaways

  • ▸Deleted Google API keys remain usable for up to 23 minutes (averaging 16 minutes) due to inconsistent revocation propagation across Google's infrastructure
  • ▸Attackers can extract files, exfiltrate cached Gemini conversations, and incur unauthorized charges during the revocation window
  • ▸Google's automatic billing tier upgrades can escalate potential damages from thousands to tens of thousands of dollars without developer consent
Source:
Hacker Newshttps://www.theregister.com/devops/2026/05/21/threat-hunters-find-google-api-keys-still-usable-23-minutes-after-deletion/5244504↗

Summary

Security researchers at Aikido have disclosed a significant vulnerability in Google's API key revocation system, finding that deleted API keys can remain functional for up to 23 minutes as the revocation propagates unevenly across Google's infrastructure. During this window, attackers with access to a leaked key can extract sensitive files from Google Gemini projects, exfiltrate cached conversations, and rack up unauthorized charges before victims realize their credential is still active. The vulnerability is particularly dangerous in combination with Google's automatic billing tier upgrades, which can escalate charges from $250 to $100,000 for long-standing accounts when usage spikes unexpectedly. In Aikido's testing, the window averaged 16 minutes and stretched to 23 minutes at worst, with an unpredictable success rate for malicious requests—some periods showed over 90% authentication while others dropped below 1%.

  • Success rate for malicious API requests varies wildly during the vulnerability window, making it impossible for developers to know when their account is truly protected

Editorial Opinion

This research exposes a troubling mismatch between what developers expect when they delete an API key and what actually happens in Google's infrastructure. The 23-minute vulnerability window is concerning enough, but combined with automatic billing tier escalation that can increase limits from $250 to $100,000, it creates an asymmetric risk scenario that heavily favors attackers during credential compromises. Google must prioritize consistent, immediate revocation propagation across all systems and reconsider billing policies that automatically increase limits without explicit developer consent.

MLOps & InfrastructureCybersecurityAI Safety & AlignmentPrivacy & Data

More from Google / Alphabet

Google / AlphabetGoogle / Alphabet
UPDATE

Google Releases Genkit Middleware for Production-Ready Agentic Applications

2026-05-21
Google / AlphabetGoogle / Alphabet
INDUSTRY REPORT

Google's Compute Crunch Drives Top AI Researchers to Launch Startups

2026-05-21
Google / AlphabetGoogle / Alphabet
INDUSTRY REPORT

Gemini's Production Disaster Exposes Dangers of 'Vibe Coding' as AI Agents Go Rogue

2026-05-21

Comments

Suggested

Google / AlphabetGoogle / Alphabet
UPDATE

Google Releases Genkit Middleware for Production-Ready Agentic Applications

2026-05-21
HiCloud TechnologyHiCloud Technology
PRODUCT LAUNCH

China Launches World's First Commercial Offshore Wind-Powered Underwater Data Center

2026-05-21
Lambda LabsLambda Labs
PARTNERSHIP

Lambda Partners with Hudson River Trading to Accelerate Quantitative Research with NVIDIA HGX B200

2026-05-21
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us