Google’s Gemini Billing Controversy Shows Why AI Infrastructure Risks Matter

[cryptoenthusiast]

Several developers recently reported receiving massive unexpected Google Cloud bills after attackers exploited API keys tied to Gemini AI services. In some cases, developers claimed they never intentionally enabled Gemini access, yet compromised API keys generated tens of thousands of dollars in usage charges within minutes.

Reported incidents included more than $10,000 in charges generated within roughly 30 minutes, while another developer faced approximately AUD $17,000 in unauthorized usage. Researchers also noted that billing limits may have been automatically increased based on account activity, potentially exposing users to significantly larger financial risks than expected.

Security researchers additionally discovered that revoked API keys could reportedly remain usable for up to 23 minutes due to delayed revocation propagation inside Google’s infrastructure. During that period, attackers may still access files, prompts, or cached AI conversations.

As AI services become deeply integrated into cloud platforms, developers and startups are increasingly learning that operational security failures can quickly escalate into major financial liabilities.

[nihalsari]

the future apparently includes devices recording your conversations while attackers speedrun five figure cloud bills in the background