Google is facing developer backlash after restricting access to its new Antigravity “vibe coding” platform for users employing the open-source AI agent OpenClaw. The move, enacted on Monday, February 23, 2026, stems from allegations of “malicious usage” that Google says overwhelmed the system and degraded service for other users. The crackdown highlights a growing tension between the flexibility of open-source AI tools and the control sought by major tech companies over their powerful models.
Several developers reported losing access to their Google accounts after using OpenClaw in conjunction with Antigravity, or connecting OpenClaw agents to their Gmail accounts. Google contends that these users were leveraging Antigravity to access a disproportionately large number of Gemini tokens through third-party platforms like OpenClaw, straining the system’s resources. This incident underscores the architectural challenges and trust concerns inherent in allowing external agents to interact with proprietary AI infrastructure.
The timing of Google’s action is particularly noteworthy, coming just one week after OpenAI announced that Peter Steinberger, the creator of OpenClaw, had joined the company to lead its “next generation of personal agents.” While OpenClaw remains an open-source project, Steinberger’s move to OpenAI – a direct competitor to Google – adds another layer of complexity to the situation. By limiting OpenClaw’s access to Antigravity, Google is effectively cutting off a pipeline that allowed an OpenAI-adjacent tool to utilize its advanced Gemini models.
A Shift in Control: The “Walled Garden” Approach
Google DeepMind engineer Varun Mohan explained the decision in a post on X (formerly Twitter), stating that the company observed “malicious usage” that led to significant service degradation. “We’ve been seeing a massive increase in malicious usage of the Antigravity backend that has tremendously degraded the quality of service for our users,” Mohan wrote. He added that Google is working to restore access for users who were unaware of the policy violation, but acknowledged limited capacity.
A Google DeepMind spokesperson clarified to VentureBeat that the move isn’t intended as a permanent ban on third-party platform access, but rather an effort to align usage with the platform’s terms of service. However, the incident has sparked a wider conversation about the future of AI agent interoperability and the increasing trend toward “walled garden” ecosystems. Anthropic took a similar step earlier this year, implementing “client fingerprinting” to restrict access to its Claude Code environment and prevent use with third-party wrappers like OpenClaw.
OpenClaw’s Evolution and Enterprise Implications
OpenClaw gained traction as a tool allowing users to run shell commands and access local files, fulfilling a key promise of AI agents: streamlining workflows. However, its open-source nature and rapid evolution have similarly presented security and governance challenges. While companies are developing secure, enterprise-grade access to OpenClaw, the platform remains relatively new, and further development is expected.
The current situation highlights the uncertainty surrounding the integration of tools like OpenClaw into existing workflows. Google’s response wasn’t framed as a security issue, but rather as a matter of access and resource management. This suggests that even with robust security measures, the sheer scale of requests generated by agents like OpenClaw can pose operational challenges for AI providers.
What This Means for Developers and Enterprises
The “Antigravity Ban” serves as a case study in the risks of relying on third-party agent integrations. For enterprise technical decision-makers, several key takeaways emerge. First, platform fragility is now a significant concern. Even high-paying customers have limited leverage when providers change their “fair use” definitions. Second, enterprises should prioritize agent frameworks that can run “local-first” or within Virtual Private Clouds (VPCs) to reduce dependency on external services. Finally, decoupling AI development from core corporate identity systems is crucial to avoid widespread disruptions from a single Terms of Service violation.
OpenClaw creator Peter Steinberger has announced that the project will remove Google support in response to the restrictions. Some users have already stated they will no longer use Google or Gemini for their projects. For those who wish to continue using Antigravity, they will necessitate to wait for Google to establish a method for OpenClaw integration that aligns with its policies.
the incident signals a shift in the AI landscape. As Google and OpenAI solidify their positions, enterprises must weigh the benefits of open-source interoperability against the stability and control offered by vertically integrated ecosystems.
Google DeepMind reiterated that access was only cut to Antigravity, and not other Google applications.
The next step in this evolving situation will be Google’s communication regarding a potential path forward for OpenClaw integration, or a clarification of its terms of service. Developers and enterprises will be closely watching for updates as the industry navigates the challenges of balancing innovation with control in the age of AI agents.
Have thoughts on Google’s decision or the future of AI agents? Share your comments below.
