It started with a brief scare.

Early Friday, OpenClaw creator Peter Steinberger shared that his Anthropic account had suddenly been suspended. The reason given was “suspicious” activity, and he posted a screenshot on X to show what had happened.

But the issue didn’t last long.

Within a few hours, after his post began getting attention online, his account was restored. The situation quickly turned into a wider conversation, with many people reacting and trying to understand what went wrong. An Anthropic engineer even stepped in publicly, saying the company doesn’t ban users just for using OpenClaw and offering to help resolve the issue.

Even now, it’s not fully clear what triggered the suspension or what led to the reversal. Still, the incident has drawn attention to a bigger shift happening behind the scenes.

Recently, Anthropic changed how its Claude subscriptions work. Third-party tools like OpenClaw are no longer included in standard plans. Instead, users now have to pay separately through API usage, depending on how much they run these tools.

The company says this is because tools like OpenClaw use significantly more computing power. They don’t just send simple prompts—they can run ongoing reasoning loops, retry tasks automatically, and connect with multiple external services, which increases load on the system.

Steinberger, however, seemed skeptical about that explanation. He pointed out that the timing feels unusual, suggesting that similar features had just been added to Anthropic’s own tools before restrictions on open-source alternatives came in. While he didn’t name them directly, the comment appeared to reference recent updates to Anthropic’s in-house agent features.

His frustration was also visible in how he responded to people online. When someone suggested he had made the wrong choice by joining OpenAI instead of Anthropic, he replied sharply, saying one company welcomed him while the other responded with legal threats.

At the same time, he made it clear that he still uses Claude—but only for testing. His goal is to make sure OpenClaw continues to work properly across different AI platforms. He also explained that his work with OpenClaw is separate from his role at OpenAI, where he focuses on future product strategy.

One interesting detail that came up during the discussion is that many OpenClaw users still prefer Claude over other models. Steinberger acknowledged that and hinted that improvements may already be in progress.

For now, the episode may have been resolved quickly, but it has raised larger questions about how open-source tools and major AI platforms will coexist going forward.

Also Read: Anthropic Accidentally Exposes Claude Code Source Code