Google bans Antigravity access for OpenClaw users, Peter Steinberger plans to drop support

Peter Steinberger, the creator of OpenClaw, called the Google ban wave “pretty draconian”. He said he will remove support for Antigravity from the tool and warned users to be careful if they plug it in.

According to reports online, the problem seems to have started for many paying subscribers in the last week or so. Affected users used Google AI Pro or Ultra plans, and they connect those accounts through OpenClaw or tools like OpenCode. Then, out of the blue, Google detected the usage and disabled Antigravity access.

Users, including paying $250/month Google AI Ultra subscribers, are seeing the following notice: “This service has been disabled in this account for violation of Terms of Service,” with no warning email, no grace period, and no obvious appeal path. Reports of mass 403 PERMISSION_DENIED errors flooded Google’s support forums and Reddit starting around February 12–14, all tied to the same root cause: using Antigravity OAuth tokens inside OpenClaw.

The core issue is a policy violation regarding “malicious usage.” Google’s Antigravity backend is intended for its own first-party services. When OpenClaw or other third-party tools route requests through these private OAuth tokens, Google’s systems flag it as abuse or circumvention of their standard API limits. The worst part is that users lose access to the entire account. Yes, YouTube and all included.

For what it’s worth, Google’s side of the story has now surfaced too. Varun Mohan, responding on X, said they had been “seeing a massive increase in malicious usage of the Antigravity backend that has tremendously degraded the quality of service” for legitimate users, adding that they needed to “quickly shut off access to these users that are not using the product as intended.” He also noted there is a path for unaware users to potentially regain access.

The topic is now trending on X, which gives you a sense of how much traction this has picked up.

This doesn’t seem to be just a Google problem, though. From what I could gather, it’s part of a wider turf war playing out across the AI coding tool space. Yuchen Jin, who posted about the broader trend on X, pointed out that “the coding LLM war escalated after OpenAI acquired the creator of OpenClaw” and that both Anthropic and Google have now blocked OpenCode from using their Pro plan subscriptions, leaving only OpenAI openly accessible.

Anthropic had already moved first back in late January, implementing client fingerprinting to block OpenCode and other third-party tools from piggybacking on Claude Code subscriptions. We covered Anthropic’s side of that story in detail here.

It does seem like the bigger labs are cracking down on third-party tools that tap into their subsidized consumer plans as a cheaper route to premium model access. Interestingly, OpenAI appears to be the odd one out here. They officially whitelisted OpenCode, and then went ahead and acquired OpenClaw itself, bringing Steinberger on board to lead personal agents development.

For now, if you’re an OpenClaw user relying on Antigravity, the community’s advice is to just stop. OpenClaw’s own documentation doesn’t list Google Antigravity as a supported provider, and the official recommendation is to use API keys rather than OAuth tokens. Alternatives like MiniMax and OpenAI’s Codex are being floated as safer options that won’t put your Google account at risk.

We stand out from the tech-media crowd because we break news stories; we mainly bring you stuff that you won’t find anywhere in the mainstream tech media. Our stories have been picked up by some of the world’s most popular websites and media outlets—more info is available here.