Source: Xinzhiyuan
Just now, the AI community witnessed a historic event of "closing the country to the outside world".
Anthropic has officially prohibited the use of its own plans to access OpenClaw!!!
Boris Cherny, the father of Claude Code, announced:
Starting from 3:00 PM Eastern Time on April 4th (3:00 AM Beijing Time on April 5th), Claude will block all third-party tools; these tools can only be used with additional plans or via API.
This means that thousands of developers and startups relying on OpenClaw to boost efficiency have overnight lost the benefit of "unlimited" usage, forced to switch to an extremely expensive "pay-as-you-go" model.
Loyal Claude users on OpenClaw have been dealt a heavy blow.
The timing of this announcement is also very telling—OpenClaw's creator, Peter Steinberger, recently moved to OpenAI, and Anthropic's intentions are clear!
This is nothing but a commercial revenge disguised as policy.
The news quickly topped the Hacker News developer community.
Let us remember this day: April 4, 2026. From this day on, the AI industry moves from open collaboration to巨头割据 (giant fragmentation).
An official Claude email proves that this time, Anthropic is specifically targeting OpenClaw!
This policy will be enforced first on OpenClaw starting April 4th, but applies to all third-party toolchains and will soon be extended to more tools.
As a consolation, Anthropic offered a one-time subsidy, exactly equal to one month's subscription fee. Valid until April 17th.
The Hammer Falls
Note, the timing of Anthropic's email was rather unbecoming.
April 3rd, Friday evening, the favorite time for internet companies to release bad news.
Anthropic notified thousands of OpenClaw users: starting tomorrow, your Claude subscription quota can no longer be used for OpenClaw; if you want to continue, you must pay as you go.
https://x.com/VadimStrizheus/status/2040199979927482618
Three months of siege finally ended with the last blow.
The Trigger: A "Defection" That Could Change the Landscape
Why would Anthropic choose this moment to ruthlessly attack an open-source tool, disregarding appearances?
Because the soul of OpenClaw, its creator Peter Steinberger, joined their "archenemy" OpenAI.
Once, Peter Steinberger was one of the developers who understood the Claude ecosystem best; his OpenClaw made Claude incredibly easy to use.
Now, to Anthropic, OpenClaw has become an enemy "Trojan horse".
Anthropic believes OpenClaw is no longer a pure efficiency tool, but an "intelligence gatherer" reaching into their backyard.
Since the founder is now with OpenAI, your tool can no longer freeload on our subscription quota.
Peter himself spoke out helplessly, hinting that Anthropic is "locking the door and beating the dog," "freeloading" on the open-source community:
Dave Morin (OpenClaw board member) and I tried to persuade Anthropic to stay calm.
But all we could achieve was to delay this day by one week.
https://x.com/steipete/status/2040209434019082522
Developers in Dire Straits, Budgets Overnight
For ordinary developers, this ban is a "dimensional打击" (devastating blow).
Previously, many developers used Claude's fixed monthly subscription, combined with OpenClaw's powerful interface, to achieve extremely low-cost automated workflows.
Spending $20 on a Claude Pro plan allowed the lobster (OpenClaw) to call Claude 24/7. The same usage through the API channel could run up a bill of thousands of dollars.
One is a $200 capped Max subscription, the other is a four-figure API cost.
Now, this path has been cut off by Anthropic.
Pay-as-you-go means no longer a flat monthly fee, making costs highly unpredictable. Many small and medium-sized teams had fixed AI budgets, which could now be exceeded at any time.
Even worse, if you don't want to pay this exorbitant toll, you must painfully refactor your entire business logic within 24 hours.
One word—Ruthless!
The Lobster Stirred the Pot, Grudges Started Long Ago
The grudge between the Lobster's father and Anthropic started long ago.
Steinberger once publicly complained that Anthropic's dealings with him were "mostly through lawyer letters".
First move, brand separation.
End of January, a lawyer letter forced Clawdbot to change its name.
Second move, technical blockade.
January 9th, Anthropic quietly added a server-side check: subscription tokens not issued from the official Claude Code client were directly rejected.
OpenClaw's core玩法 (play/gimmick) was reset overnight.
Third move, terms定性 (qualification/definition).
Mid-February, service terms updated: Using OAuth tokens from Free, Pro, Max accounts in any third-party tool is a violation.
The fiercest fourth move was feature copying.
Claude Cowork launched Dispatch, allowing remote control of desktop Claude from a phone; Claude Code launched Channels, integrating with Telegram and Discord.
Within four weeks, OpenClaw's core features were one-to-one replicated by the official team!
As AI blogger Matthew Berman put it: "They basically built their own OpenClaw."
Tech media Semafor reported this morning that when Anthropic's Chief Business Officer Paul Smith was asked if customers were asking the company to build its own OpenClaw, he admitted they were.
https://www.semafor.com/article/04/03/2026/anthropic-eyes-its-own-version-of-openclaw
And today, April 4th's email, is the final blow.
Open Scheme: Pushing Their Own "Favorite Son"
Claude Cowork
While banning OpenClaw, Anthropic also showed its true colors.
They began heavily暗示 (hinting): Stop using those unreliable third-party tools, come try our native Claude Cowork!
Claude Cowork allows Claude to control the coding environment and computer interface more deeply.
Actually, this isn't a unique play for Anthropic either.
This is the巨头们 (giants') most adept "platform lock-in" strategy:
-
Step One: Use third-party open-source tools to attract developers and grow the ecosystem.
-
Step Two: Find an excuse (e.g., security, infrastructure pressure) to ban third parties.
-
Step Three: Force users to migrate to their own native integration tools which are more expensive and tightly controlled.
This typical "vertical integration" tactic allows them to firmly control the entry point and user experience, while gradually diminishing the advantages of those "uncontrolled connectors" in the ecosystem.
Interestingly, OpenAI chose a completely opposite path almost simultaneously.
OpenAI explicitly allows Codex subscriptions to be used in third-party clients like OpenClaw.
In March, they went further, announcing free ChatGPT Pro access for open-source project maintainers, with OpenClaw specifically named on the beneficiary list. The people Anthropic wants to drive away, OpenAI welcomes them all!
No One is Completely Innocent
To be fair, Anthropic has its reasons.
Every user who uses a $200 subscription to generate thousands of dollars worth of API usage is causing the company to lose money.
Third-party tools bypass official telemetry, disguise client identities, and create monitoring blind spots—these are real engineering and security risks.
Especially since Steinberger now sits in the OpenAI office—OpenClaw, in Anthropic's eyes, has gone from "an annoying freeloader" to "a spy sent by the other side".
Moreover, OpenClaw itself isn't clean.
It was exposed to a high-risk vulnerability (CVE-2026-25253) with a CVSS score of 8.8, allowing attackers to steal user authentication tokens via a link.
https://nvd.nist.gov/vuln/detail/CVE-2026-25253
On the public internet, security agencies have scanned over thirty thousand wide-open OpenClaw instances.
But these justifications don't dissolve the developers' feeling of being played.
The platform first opens the door, lets people come in and build, then announces different plans—this script has been played too many times.
X killing third-party clients, Apple tightening the App Store, Google cutting free APIs, are all the same story.
Every time, it's the developers who get hurt.
The Open Ecosystem Enters Twilight, AI Giants Close Their Doors
Today's ultimate move by Anthropic makes us understand a heartbreaking fact—
The golden age of AI, which belonged to developers and was free and open, is coming to an end.
Once, we thought AI would be like the early internet, based on shared protocols and community evolution. But the reality is, model owners hold the power of life and death, able to nullify the efforts of thousands at any time.
Now, the countdown to the 3:00 PM April 4th ultimatum has begun.
Developers face three harsh choices—
Swallow the bitter pill, endure sky-high pay-as-you-go fees, continue using OpenClaw until the budget runs out?
Or fully migrate, kneel before Anthropic's native tools, and accept platform lock-in?
Or, leave in anger, completely抛弃 (abandon) Claude.
Finally, while the giants are busy fighting among themselves, please don't forget, it was the thousands of developers with their lines of code who built your billion-dollar valuation thrones.
This time, Anthropic won the competition, but彻底输掉了 (completely lost) the trust of the community!
