Why did Google restrict Antigravity users?

What triggered Google’s enforcement and what changed

Google moved to limit and, in some cases, suspend customer access to Antigravity after a wave of agent-driven activity began consuming large amounts of backend compute. That surge was tied to third‑party agent frameworks that automate long-running or massively parallel prompts against Gemini models. Those agentic workflows produced unexpected sustained load on Google’s systems, prompting the company to label certain usage patterns as "malicious" and to cut off accounts that were linked to the activity.

The incident exposed a basic capacity and policy gap: developer-built agents can orchestrate rapid, continuous calls in ways that the standard API and subscription rules didn’t anticipate. Paid Pro and Ultra-tier customers reported bans after hooking agent runtimes into Antigravity; some developers said they were forced to remove or rethink Gemini integrations to avoid having users blocked.

What this means for developers and enterprises

• Expect stricter rate limits and new acceptable‑use controls on agent-style workloads.
• Operators of agent frameworks should add client-side safeguards: backoffs, token budgets, and identity attestations.
• Customers running automated workflows should monitor unit costs and system telemetry closely to avoid abrupt service interruptions.

Google framed the move as protecting capacity and preventing abuse. For developers who built or relied on agentic tooling, the episode is a reminder that autonomous, closed‑loop systems can rapidly outstrip the guardrails of commercial APIs. The situation remains in flux: some platform maintainers are debating whether to permanently remove integrations, while users press vendors for clearer rules and more predictable limits.