Recently, a substantial number of paid subscribers to Google's AI agent development platform, Antigravity, have reported experiencing account bans—and, in some cases, the deactivation of their linked email services—following the integration of third-party AI coding tools like OpenClaw and OpenCode. Antigravity, an AI-powered integrated development environment (IDE) platform launched by Google in November 2025, incorporates agent capabilities and can access Google's foundational models. The wave of bans commenced around February 12, 2026, with certain users initially encountering 403 permission errors. The centralized enforcement phase took place on February 23-24 without any prior notification, impacting a broad spectrum of users, including those subscribed to the AI Ultra paid plan, which costs $250 per month. Some accounts also saw disruptions to core associated services such as Gmail and Google Workspace.

Google DeepMind engineer Varun Mohan attributed the bans to "malicious use" by OpenClaw users, noting that a surge in illicit invocations of Antigravity's backend had been detected. A significant number of users were leveraging OpenClaw to bypass official interfaces, automating high-frequency concurrent calls to the Gemini model. This practice far exceeded the platform's intended quota for human-interactive computing power, severely degrading service quality for legitimate users.

However, affected users have criticized the severity of Google's enforcement measures, with some accounts being permanently frozen and no refunds offered to those who had prepaid annual fees. The lack of prior warning and absence of appeal channels have sparked widespread controversy.

This incident coincides with the news that OpenClaw founder Peter Steinberger has joined OpenAI. He responded vehemently to Google's bans, labeling them as "extremely harsh" and announcing that future versions of OpenClaw would completely terminate support for Google's Antigravity and Gemini, severing all ties with Google's ecosystem. Notably, just three days prior to the bans, OpenClaw had officially announced support for the Gemini 3.1 Pro model.

Industry analysts suggest that while Google's stated rationale for the bans is to curb computing power abuse and "token arbitrage," the underlying motive appears to be a fierce struggle for control over the AI ecosystem. This incident is not an isolated case but rather reflects a broader trend of AI giants tightening their ecological control, marking a further acceleration in the industry's shift from "open-source sharing" to "closed silos."