The Ban That Lasted Hours But Spoke Volumes
Peter Steinberger did everything right.
The creator of OpenClaw-an open-source AI assistant that lets you run autonomous agents across WhatsApp, Telegram, Slack, and iMessage-received an email from Anthropic on Friday morning. His account had been suspended. Reason: “suspicious activity.”
He shared the screenshot on X. Within hours, the post went viral. And just as quickly, his access was restored.
But the damage was already done.
This wasn’t just a routine account flag. It was the latest salvo in a growing war between Anthropic and the open-source developer community-specifically, anyone building third-party tools around its Claude AI models.
And for developers in India who rely on OpenClaw to automate workflows, this is a wake-up call: the era of cheap, unrestricted AI access is ending. And the “claw tax” is just the beginning.
The Timeline-How We Got Here
To understand the ban, you need to understand the week that led up to it.
April 4, 2026: Anthropic announced that Claude subscriptions would no longer cover third-party tools like OpenClaw. Users would now have to pay separately through the API on a consumption basis. The company called it a necessary move to manage “usage patterns” that subscriptions weren’t built for. The community called it something else: a “claw tax.”
April 10, 2026 (Morning): Steinberger, who now works at OpenAI but continues to maintain OpenClaw as an open-source project, was testing Claude compatibility to ensure updates wouldn’t break for his users. He received a suspension notice. “It’s gonna be harder in the future to ensure OpenClaw still works with Anthropic models,” he posted.
April 10, 2026 (Afternoon): After his post gained traction, an Anthropic engineer publicly responded, saying the company had “never banned anyone for using OpenClaw” and offered to help. Hours later, Steinberger’s account was reinstated.
Steinberger’s own account of what triggered the ban? He had been working on getting a Claude fallback feature working-something an Anthropic employee had confirmed was a classifier bug. That work, he said, “seems to have gotten me banned too.”
The “Claw Tax”-What It Really Costs You
Let’s talk numbers, because this is where the story hits home for Indian developers and businesses.
Before April 4, a $20 Claude Pro subscription gave you generous usage limits. OpenClaw tapped into those limits directly. You paid one flat fee and could run autonomous agents-clearing inboxes, managing calendars, checking flight statuses, sending reminders-to your heart’s content.
After April 4, OpenClaw usage is decoupled from your subscription. You pay separate pay-as-you-go rates for every API call.
What does that look like in practice?
One commentator put the burn rate for a single autonomous agent at $1,000 to $5,000 a day in API fees-costs that Anthropic had previously absorbed when usage ran through subscriptions instead of metered billing.
That’s not a typo. A single OpenClaw agent running continuous reasoning loops could now cost you more than a month’s rent in a single day.
Anthropic is offering a one-time credit equal to your monthly plan cost and discounted usage bundles. But that’s a temporary bandage, not a fix.
Read alos: Claude vs. ChatGPT vs. Gemini: The Winner Isn't Who You Think
The “Copy and Lock Out” Strategy
Steinberger’s frustration boiled over in a post that cut straight to the heart of the matter: “Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source.”
What features? Two weeks before the pricing change, Anthropic rolled out “Claude Dispatch” for its own agent, Cowork-allowing users to remotely control agents and assign tasks.
Sound familiar? It should. Those are core capabilities that OpenClaw users have had for months.
The pattern is unmistakable: observe what the open-source community builds, replicate it in your proprietary product, then change the rules to make the original harder to use.
When someone on X suggested Steinberger should have taken a job at Anthropic instead of OpenAI, his response was brutal: “One welcomed me, one sent legal threats.”
Why Steinberger Still Uses Claude (And Why That Matters)
Here’s the twist that makes this story even stranger.
Steinberger now works at OpenAI. Yet he was testing Claude. Why?
Because OpenClaw is designed to work with any model provider. Steinberger’s role at the OpenClaw Foundation is to ensure compatibility across the board-not to push OpenAI’s models. He uses Claude strictly for testing, to make sure updates don’t break for OpenClaw users who prefer Anthropic.
“You need to separate two things,” he explained. “My work at the OpenClaw Foundation where we wanna make OpenClaw work great for any model provider, and my job at OpenAI to help them with future product strategy.”
That distinction was lost on Anthropic’s automated systems-and, some would argue, on Anthropic’s policy team as well.
What This Means for Indian Developers
If you’re using OpenClaw with Claude in India, here’s your new reality:
Your costs are about to spike. If you were relying on a Claude subscription to power your automations, you’ll need to budget for API usage instead. For heavy users, the difference could be orders of magnitude.
Your workflow is less stable. Anthropic has now demonstrated that it can and will change access rules with little notice. Building a business on top of a third-party AI model carries platform risk. That risk just became very real.
Alternatives exist-but they come with trade-offs. You could switch to OpenAI’s models, but they have their own usage policies. You could use Anthropic’s own Cowork, but it’s not open source and doesn’t have OpenClaw’s multi-platform reach. Or you could self-host with smaller models, but that requires significant technical expertise.
One workaround gaining traction: using API keys from aggregators like OpenRouter, which abstract away the provider and let you switch models without changing your code. But even those come with their own costs and rate limits.
Read also: Humanoid Robots Just Went from Lab Demo to National Spotlight
The Bigger Picture-AI’s Walled Garden Moment
The OpenClaw ban isn’t an isolated incident. It’s part of a pattern.
Anthropic previously cut off OpenAI’s access to Claude’s API, citing competitive benchmarking. It later blocked Cursor from using Claude models. And now, OpenClaw is being pushed to pay-as-you-go.
The message is consistent: Anthropic is building a wall around its garden. If you want to play inside, you pay by the hour-and you don’t bring your own tools.
For the open-source community, this is a betrayal of the values that built the AI boom. For Anthropic, it’s a necessary step to manage costs and prioritize paying customers.
For developers in India, it’s a reminder: never build your house on rented land.
What You Should Do Right Now
If OpenClaw is critical to your workflow, here’s your action plan:
1. Audit your usage. How many API calls are you making? How much would that cost under the new API pricing? Run the numbers before you get surprised.
2. Consider alternatives. OpenClaw works with OpenAI, Google Gemini, xAI’s Grok, and self-hosted models like Llama and Mistral. Diversifying your provider reduces platform risk.
3. Buy usage bundles now. Anthropic is offering discounted bundles as a transitional measure. If you’re staying with Claude, stock up while the discount lasts.
4. Watch the legal landscape. The OpenClaw Foundation is exploring legal options around platform lock-in and anti-competitive behavior. This story isn’t over.
Conclusion: The Party’s Over
For two years, AI enthusiasts enjoyed an all-you-can-eat buffet. A $20 subscription gave you unlimited access to frontier models, and open-source tools like OpenClaw turned those models into autonomous workers that never slept.
That era is ending.
Anthropic’s “claw tax” is the first major pricing correction in the AI industry. It won’t be the last. OpenAI, Google, and others are watching closely-and taking notes.
The question isn’t whether your AI costs will rise. It’s by how much-and whether you’ll see it coming.
Read also: The Government Just Handed Coinbase a Weapon. Traditional Banks Are Terrified.
Share This With Every Developer You Know
Tag a colleague who uses OpenClaw. Share this in your tech WhatsApp group. Post it on LinkedIn with the caption: “The AI all-you-can-eat buffet is over. Here’s what you need to know.”
The era of cheap AI agents is ending. Don’t get caught off guard.
FAQ
Q: Is Anthropic banning OpenClaw completely?
A: No. It’s a financial restriction, not a technical block. You can still use OpenClaw with Claude, but you’ll have to pay separately through the API on a consumption basis. Your subscription no longer covers it.
Q: Why did Anthropic change the policy?
A: Anthropic claims that subscriptions weren’t built for the “usage patterns” of autonomous agents like OpenClaw, which run continuous reasoning loops and automated retries-far more compute-intensive than normal chat.
Q: How much more will I pay?
A: It depends entirely on your usage. One commentator estimated a single agent could cost $1,000–$5,000 per day in API fees. Heavy automation users will see costs multiply significantly.
Q: Is this related to Steinberger working at OpenAI?
A: Anthropic says no. An engineer publicly stated the company has “never banned anyone for using OpenClaw.” However, the timing-just a week after the pricing change-has led to widespread speculation.
Q: What are my alternatives to Claude with OpenClaw?
A: OpenClaw supports OpenAI, Google Gemini, xAI’s Grok, and self-hosted models like Llama and Mistral. Switching providers is possible, though each has its own pricing and limitations.
Have a question about AI or the latest tech trends? We’d love to hear your thoughts!
Please stay on topic and keep it helpful. Note: All comments are moderated to keep our community spam-free.