Google Bans Hundreds of Paying Antigravity Users for Using OpenClaw - Then Says "We Heard You"
The Google Antigravity OpenClaw controversy erupted in mid-February 2026, when numerous users of Google’s agentic development platform Antigravity (powered by Gemini models) suddenly found their accounts restricted or banned. The trigger: integration with OpenClaw, the viral open-source AI agent framework that exploded in popularity late 2025.
What Happened: The Ban Wave
Around February 12–14, 2026, reports flooded forums like the Google AI Developers Forum, Reddit, and X. Paid subscribers, including those on the pricey AI Ultra plan ($250/month), lost access without warning, no prior notice, and initially no clear appeals process. Google cited “malicious usage” and backend overload from automated/agentic patterns routed through third-party tools via OAuth credentials.
Many affected users weren’t intentionally abusing quotas; they used OpenClaw to power advanced workflows (e.g., agent chains, automated coding tasks) that leveraged Antigravity’s flat-rate access instead of metered APIs. This created massive token consumption spikes, degrading service for standard users and straining Google’s infrastructure.
The official response came on February 27, 2026, from the @antigravity account:
To the builders: we heard you. We're welcoming back everyone who recently had their Google Antigravity accounts restricted for use of third-party tools. Moving forward, we’ll have clear steps for users to restore their account if it’s restricted.
— Google Antigravity (@antigravity) February 27, 2026
To maintain the integrity of…
This post confirmed:
- Mass restoration for those restricted.
- A promised future restoration process (though with limited capacity mentioned in related coverage).
- Strict enforcement: third-party tool use via Antigravity login stays prohibited under ToS (likely under clauses banning credential sharing, unauthorized product powering, or resource abuse).
OpenClaw Founder’s Reaction and Support Removal
OpenClaw’s creator, Peter Steinberger (@steipete), was vocal and critical. He called Google’s approach “pretty draconian” in posts around February 23, 2026, contrasting it with more communicative providers like Anthropic:
Pretty draconian from Google. Be careful out there if you use Antigravity. I guess I'll remove support.
— Peter Steinberger 🦖 (@steipete) February 23, 2026
Even Anthropic pings me and is nice about issues. Google just... bans? https://t.co/JBq9YCB7nB
He followed through: OpenClaw removed Antigravity support in updates around late February 2026. Community reports and coverage confirm the OAuth bridge was moved to a bundled (disabled-by-default) plugin, and official docs no longer list Google Antigravity as a supported provider. Users updating to versions like 2026.2.23 found the integration gone or requiring manual re-enablement that was discouraged.
This move came amid irony - Steinberger had recently joined OpenAI (announced mid-February 2026) to lead next-gen personal agents, with OpenClaw continuing as an open-source project under OpenAI support.
Timeline Summary
- Early–mid February 2026: Surge in OpenClaw + Antigravity usage → backend strain.
- Feb 12–14: Mass restrictions begin; complaints spike on forums/Reddit/X.
- Feb 15: OpenAI hires Steinberger.
- Feb 23: Steinberger publicly criticizes Google as “draconian” and announces intent to drop support.
- Late Feb: OpenClaw updates remove/disable Antigravity integration.
- Feb 27: Google @antigravity announces restorations and reaffirms policy.
The Way Forward for Antigravity Users
- Stick to official channels - Use the native Antigravity IDE, Gemini CLI (where permitted), and avoid third-party OAuth routing to prevent restrictions.
- If restricted - Follow the upcoming restoration steps outlined by Google (check @antigravity or discuss.ai.google.dev for updates).
- Alternatives for agentic/power users:
- Local/open models (e.g., Gemma, Qwen on hardware).
- Providers with clearer agent-friendly rules (MiniMax, Anthropic with direct APIs).
- Self-hosted agents or metered API usage to avoid flat-rate credential risks.
- Broader lesson - Flat-rate AI access + open agent tools create inevitable tension. Expect more providers to tighten ToS, add explicit prohibitions, or launch dedicated agent APIs.
The incident highlights growing pains in agentic AI: innovation thrives on openness, but scale demands guardrails. Google’s reversal shows they listen to backlash - but the firm line on third-party tools signals the “wild west” era may be closing.
Sources: Official @antigravity announcement, coverage from The Register, MLQ.ai, PCWorld, Google AI forum threads, and OpenClaw GitHub/community updates.
OpenClaw Framework
The viral open-source AI agent framework that led to the Antigravity ban wave.
View on GitHub