Google has restricted certain subscribers of its premium Gemini AI Ultra plan after detecting access routed through OpenClaw, an open-source agent framework. The enforcement has locked some users out of Google’s AI coding product, Antigravity, and sparked debate about how AI companies regulate subscription usage.
Gemini AI Ultra Users Report Sudden Account Restrictions
Customers subscribed to the $249.99-per-month Gemini AI Ultra plan reported losing access after connecting Gemini via OpenClaw using Antigravity OAuth tokens. Several affected users encountered 403-style “Terms of Service” errors without prior warning.
Because Google accounts often serve as centralized authentication across multiple services, restrictions tied to AI usage can potentially affect broader account functionality. This has amplified frustration among users who believed their paid subscription covered advanced usage scenarios.
Google Responds: Protecting Service Quality and Preventing Misuse
On February 23, Varun Mohan, lead of Antigravity, addressed the situation publicly. He stated that Google observed a “massive increase in malicious usage” of the Antigravity backend, which negatively impacted overall service quality.
According to Mohan, the company needed a rapid enforcement mechanism to block accounts “not using the product as intended.” He also acknowledged that some affected users may not have clearly understood the rules and indicated that Google intends to provide a path for reinstatement in certain cases.
The enforcement, however, highlights how AI providers are balancing system capacity, abuse prevention, and customer expectations.
OpenClaw OAuth Access Raises Policy Questions
OpenClaw enables users to route AI requests through third-party workflows, often increasing automation and altering usage patterns. While some Gemini AI Ultra subscribers believed their premium plan permitted such integrations, Google’s automated systems reportedly classified that access path as unauthorized.
This case underscores a growing industry-wide trend: AI providers are increasingly restricting consumer-tier subscription credentials from being used inside third-party automation frameworks. The goal is to manage system load, reduce abuse risks, and steer heavy users toward official APIs or enterprise offerings.
Subscription Access vs. Approved Usage Paths
Google markets Gemini AI Ultra as offering the “highest level of access” to its models. However, this incident illustrates a critical distinction — access depends not only on payment tier but also on how the service is used.
Flat-rate AI subscriptions can provide powerful model access, but companies may still enforce technical and policy-based limits on integration methods, automation, and backend usage patterns.
As AI agent frameworks become more common and increase request volumes, enforcement disputes like this could become more frequent across the industry.
What This Means for AI Power Users
For developers and advanced users, the situation serves as a reminder to carefully review acceptable usage policies before integrating premium AI subscriptions into third-party tools.
For AI providers, the challenge remains clear: maintain service stability and prevent abuse while ensuring paying customers understand the boundaries of permitted use.
With AI automation accelerating rapidly, the balance between open experimentation and controlled infrastructure may define the next phase of subscription-based AI services.
