Google has started banning users who connected the popular open-source AI agent OpenClaw to its Antigravity coding platform. The crackdown began on February 12, 2026, and affected hundreds of developers including paying subscribers who spent up to 250 dollars per month. The bans came without warning, without explanation, and with no path to appeal or refund.
The move highlights the growing tension between open-source AI tools and closed corporate platforms. It also raises serious questions about user rights, platform lock-in, and how much control tech giants should have over how developers use their services.
What Happened
On February 12, 2026, developers began reporting sudden account restrictions on Google’s AI Developer Forum. They had been using OpenClaw, an open-source AI agent created by Peter Steinberger, to access Google’s Gemini models through Antigravity’s OAuth authentication system. Many were paying subscribers on the AI Ultra tier at 249.99 dollars per month.
There was no advance notice. No email warning. No grace period to disconnect the tool. Users simply woke up to find their accounts suspended. Some lost access not just to Antigravity but also to Gmail, Google Drive, YouTube, and other Google services tied to the same account. Others found that their entire Google account history stretching back decades was suddenly inaccessible.
One affected user reported that they had prepaid for a full year of AI Ultra service. After the ban, they had no way to get their money back. Another user said they had only connected OpenClaw for a single day of testing before their account was permanently terminated.
Why Google Took Action
Google’s explanation came from Varun Mohan, a former Windsurf CEO who now leads Antigravity at Google DeepMind. On February 23, 2026, he posted on X that Google had detected a massive increase in what he called malicious usage of the Antigravity backend. He said this traffic had tremendously degraded service quality for legitimate users.
According to Mohan, developers were using OpenClaw to route requests through Antigravity’s servers at subsidized rates. This allowed them to access high-end models like nsfw ai art generator Gemini 2.5 Pro without paying full price. The traffic surge overwhelmed Google’s systems and affected performance for regular Antigravity users.
Google’s official position was clear. Using Antigravity servers to power a non-Antigravity product violates the Terms of Service. The company applied a zero-tolerance policy with no exceptions. Users who had their accounts suspended were told the decision could not be reversed.
However, many developers argued they had no way of knowing this was against the rules. The OAuth connection point did not display any warning about third-party tool restrictions. Google’s Terms of Service did not explicitly mention OpenClaw or similar tools at the time of the bans. Users felt they were being punished for a rule they did not know existed.
The Scale of the Damage
The ai nude generator free ban wave was not limited to a few accounts. Reports flooded developer forums, Reddit, and Hacker News for days. The pattern was consistent across all cases.
Users received a generic Terms of Service violation message with no specific explanation. Many experienced suspension of their entire Google account, not just Antigravity access. The ban appeared to be automated, triggered by detection of specific API access patterns. There was no advance warning, no grace period, and no opportunity to disconnect the tool before enforcement.
celebrity ai nudes
Some users reported that even creating fresh Google accounts for Antigravity resulted in immediate restriction. Others said they were still being charged monthly subscription fees for accounts they could not access. No automatic refunds were issued.
The human cost was real. Developers who relied on Google’s services for work, communication, and business operations found themselves suddenly cut off. One user described losing access to a Google Cloud Platform project that powered their startup. Another said they lost years of Gmail history and YouTube content.
OpenClaw’s Response
Peter Steinberger, the creator of OpenClaw, responded quickly. He called Google’s enforcement pretty draconian and announced that he would remove Antigravity support from OpenClaw entirely. He also warned users to be careful if they continue using Antigravity with third-party tools.
Steinberger noted that Anthropic had handled a similar situation differently. When Anthropic detected third-party tool usage, they reached out directly to discuss the issue. Google, by contrast, simply banned accounts without communication. Steinberger’s comment on X captured the sentiment of many developers: Even Anthropic pings me and is nice about issues. Google just bans?
The irony is that Steinberger had recently joined OpenAI as an employee. OpenAI had taken yet another approach by acquiring the team behind a similar third-party tool and whitelisting it for official use. This means OpenClaw is now backed by one of Google’s biggest competitors, making the conflict even more pointed.
How Other Companies Handled Similar Issues
The Google ban wave was not the first time an AI company cracked down on third-party tools. It was, however, the most aggressive.
Anthropic blocked OpenCode from accessing Claude using client fingerprinting techniques in late January 2026. The key difference was that Anthropic only blocked the unauthorized access method. Users lost the specific integration but kept their Claude accounts and all associated data. Anthropic also reversed some accidental bans and eventually published formal documentation explaining the policy.
OpenAI took the opposite approach. Instead of banning users, the company acquired the team behind OpenCode and made the tool an officially supported integration. This turned a potential conflict into a competitive advantage and generated positive community sentiment.
Google’s approach stood out for its severity. Instant permanent suspensions. Continued billing during suspension. Limited support recourse. No refunds for prepaid annual plans. The contrast with Anthropic’s handling of the same problem was stark.
The Broader Implications
This dispute is about more than one tool or one company. It reveals fundamental tensions in how AI platforms will operate as the industry matures.
On one side, companies like Google are building closed ecosystems. They want users to access AI models through official channels where they can control pricing, monitor usage, and optimize performance. Third-party tools that route traffic through back doors threaten this business model.
On the other side, developers want flexibility. They want to use the best tools for their workflow, regardless of which company made them. Open-source projects like OpenClaw represent this philosophy. They are model-agnostic, platform-independent, and designed to give users control.
The conflict between these two visions is inevitable. As AI becomes more central to how people work, the platforms that provide it will fight harder to keep users inside their walled gardens. The question is how aggressively they will enforce those walls, and what happens to users who try to climb over them.
What Developers Can Do
For developers who were affected by the ban wave, the options are limited but not zero.
Some users reported success in recovering their accounts by filing appeals through Google’s account recovery process. The success rate varies. Accounts with only Antigravity access suspended had about a 70 percent recovery rate within three to seven days. Full Google account suspensions were harder to reverse, with success rates around 45 to 60 percent and timelines of one to three weeks.
The most reliable path forward is to stop using subscription OAuth tokens in third-party tools. This applies to every provider, not just Google. Subscription OAuth is subsidized access, and every provider will eventually protect it. Switch to direct API keys or managed services that do not rely on platform authentication.
Open-source models offer another alternative. DeepSeek, Llama, and other open models can run on self-hosted infrastructure or through provider-agnostic services. No vendor can ban you from using an open-source model. The trade-off is slightly lower performance on some tasks, but the cost savings and freedom from platform risk are significant.
Final Thoughts
Google’s ban wave against OpenClaw users sends a clear message. The company will not tolerate third-party tools routing traffic through its subsidized access channels. The enforcement was swift, severe, and unforgiving.
For the broader developer community, the lesson is equally clear. Building on platform-specific OAuth tokens is building on borrowed time. Every major AI provider will eventually close this path because the economics demand it. Anthropic did it in January. Google did it in February. Others will follow.
The AI industry is following a predictable pattern. Open early access to build ecosystems. Then restrictions once the economics bite. The question for every developer is whether they have built on ground they control, or ground that can be pulled out without warning.
Google’s approach may protect its platform in the short term. But it also risks alienating the developer community that has made its AI tools popular. In a market where talent and trust are the most valuable assets, that is a trade-off worth watching.
