Skip to content
Submit incident
AII-1001-12231
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.Supporting sources are pending verification.

Google restricts Antigravity and Gemini users over OpenClaw OAuth abuse

Google restricted access to its Antigravity coding tool and Gemini AI models for users who connected their accounts via OpenClaw, an open-source AI agent, citing a surge in malicious usage that degraded service quality. The bans reportedly came without warning, affecting some paid Ultra subscribers, though Google later clarified that other Google services remained unaffected.

23 February 2026 Β· Google Β· Security and cyberattacks
23 February 2026
AII-1001-12231
Quick facts
Category
Security and cyberattacks
Country
United States
Date
23 February 2026
Drafted by
Raidu Incident Detection

What happened

Google restricted access to Antigravity and Gemini AI models for users who connected their accounts through OpenClaw, an open-source AI agent tool. The restrictions affected some paid Gemini AI Ultra subscribers, with some also losing access to Gmail and Google Workspace accounts.

Google DeepMind engineer Varun Mohan stated that the company detected 'a massive increase in malicious usage of the Antigravity backend that has tremendously degraded the quality of service for our users.' He noted that users accessing via OpenClaw were not using the product as intended, as this practice was against Google's terms of service.

Reported impact

Several Gemini AI Ultra subscribers reported being unable to access the Gemini 2.5 Pro model and linked services without warning or explanation. OpenClaw creator Peter Steinberger criticized the action as 'pretty draconian from Google,' noting he would likely remove support for using Antigravity OAuth credentials in OpenClaw.

Users who had been banned were unable to recover their accounts or access paid AI services. Mohan clarified that only Antigravity access was blocked, with other Google services remaining available, but acknowledged that 'we have limited capacity' to reinstate users.

Index analysis

OpenClaw reportedly consumed millions of AI tokens in a single session, overwhelming Google's compute infrastructure and degrading service quality for legitimate users. The viral AI tool's massive token consumption through flat-rate OAuth credentials prompted both Google and Anthropic to take action against users utilizing this configuration.

Governance controls

Mohan stated that Google would provide a path for users unaware of the terms-of-service violation to regain access, though he noted the company had limited capacity to reinstate accounts. Google aims to balance swift action against misuse with fairness to legitimate users.

Classification

Category
Security and cyberattacks
Industry
IT infrastructure and hosting
Country
United States
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property, Reputational
Harmed
Consumers
Failure modes
Agent misaction
Consequences
System suspension or termination
AI systems
AntiGravity, Google Gemini 2.5 Pro, OpenClaw

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident