Google has suspended access to its Antigravity AI platform for numerous users of the open-source tool OpenClaw, sparking backlash over aggressive enforcement of terms of service (ToS).
The move targets developers leveraging OpenClaw’s OAuth plugin to tap into subsidized Gemini model tokens, which caused backend spikes and service degradation.
OpenClaw, launched in November 2025, exploded in popularity with over 219,000 GitHub stars, enabling local AI agents for tasks like email management and web browsing. Users authenticated via Google’s Antigravity, its developer-facing Gemini AI platform, bypassing official channels to access high-end models like Gemini 2.5 Pro at reduced costs.
This third-party integration violated Google’s ToS by powering non-Antigravity products with Antigravity infrastructure, leading to “malicious usage” patterns flagged by automated systems.
The Suspension Wave
Reports surfaced mid-February 2026 on Google’s AI Developer Forum, where AI Ultra subscribers paying $249.99 monthly faced sudden 403 errors and full account restrictions without warnings.
Affected users lost not only Antigravity and Gemini CLI access but, in some cases, Gmail, Workspace, and even decades-old account histories, prompting accusations of overreach.
Varun Mohan, Google DeepMind product lead and former Windsurf CEO, publicly addressed the surge on X (formerly Twitter), stating it “tremendously degraded the quality of service for our users.” He emphasized prioritizing “actual users” while offering a limited reinstatement path for unaware violators, amid capacity constraints.
Just to add some clarification, we have purely blocked usage of the Antigravity product for these users. All your other Google services (and Google AI services) are unaffected. It is not intended to use the Antigravity backend as a proxy for other products and users in these…— Varun Mohan (@_mohansolo) February 23, 2026
The crackdown echoes Anthropic’s recent ToS update explicitly banning third-party OAuth in tools like OpenClaw, citing token arbitrage and debugging issues from anomalous traffic.
Cybersecurity experts highlight OpenClaw’s vulnerabilities, exacerbating the fallout: over 21,000 exposed instances vulnerable to infostealers targeting config files, plus supply chain attacks.
China’s industry ministry warned of risks from misconfigured systems that enable cyberattacks and data breaches, underscoring the double-edged sword of agentic AI. Peter Steinberger, the creator of OpenClaw who recently joined OpenAI to advance personal agents, decried Google’s “draconian” bans and announced plans to drop Antigravity support.
Pretty draconian from Google. Be careful out there if you use Antigravity. I guess I'll remove support.
Even Anthropic pings me and is nice about issues. Google just… bans? Peter Steinberger (@steipete) February 23, 2026
The community is pivoting to thriving forks like Nanobot and IronClaw, while OpenAI endorses third-party harnesses, widening competitive rifts. This turf war signals a shift to closed AI ecosystems, forcing developers to weigh convenience against ToS and security perils.
As AI agents proliferate, such enforcements remind users that subsidized access often hides backend costs, and ban waves can wipe out productivity tools overnight.
Follow us on Google News, LinkedIn, and X for daily cybersecurity updates. Contact us to feature your stories.
The post Google Suspends OpenClaw Users from Antigravity AI After OAuth Token Abuse appeared first on Cyber Security News.
