Anthropic closes door on subscription use of OpenClaw
- Reference: 1775504256
- News link: https://www.theregister.co.uk/2026/04/06/anthropic_closes_door_on_subscription/
- Source link:
Probably not because of OpenClaw, Claude was struggling on Monday with degraded service following further efforts to balance capacity with demand.
"We have identified an issue resulting in elevated errors on Claude.ai, including desktop and mobile," the company's status page said, characterizing the incident as "a partial outage." Uptime over the past 90 days slipped to 98.82 percent.
[1]
An Anthropic spokesperson did not immediately have an answer as to the source of the issue. But the disruption did not last long.
[2]
[3]
"From 15:00–16:30 UTC on April 6, we saw elevated errors on login for Claude.ai and Claude Code," the company's [4]status page said. "This issue also affected some Claude.ai conversations and other product functionality such as voice mode. This issue is now resolved."
The service disruption follows a period of high demand for Claude, one that the company has tried to address by denying access to third-party tools that use the AI service through subscriptions. On Friday, Boris Cherny, head of Claude Code, [5]said , "Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw."
[6]
Cherny explained that the restriction followed from engineering constraints. "Our systems are highly optimized for one kind of workload, and to serve as many people as possible with the most intelligent models, we are continuing to optimize that," he [7]said .
He nonetheless insisted, "We're big fans of open source. I actually just put up a few [pull requests] to improve prompt cache efficiency for OpenClaw specifically."
Google in February [8]took similar action to enforce its terms of service related to its Antigravity AI development environment, Gemini CLI, and Gemini Code Assist. "Using third-party software, tools, or services to harvest or piggyback on Gemini CLI's OAuth authentication to access our backend services is a direct violation of Gemini CLI's applicable terms and policies," said Jack Wotherspoon, Gemini CLI developer relations, at the time.
[9]
Anthropic sells AI service tokens through either subscriptions or its API. Subscribers pay a flat rate and are subject to session and monthly usage limits, with an option to pay for extra usage once capped. API customers pay per token, with no usage limits. For developers who use Claude heavily, subscription-based pricing can be significantly less expensive.
For this reporter during the month of March, a $20 monthly subscription enabled about $236 of token usage (which doesn't necessarily reflect the actual per token cost to Anthropic). Others [10]report similarly skewed ratios when price paid is compared to list price value or more – [11]36x by one measure . This would presumably be balanced somewhat by developers who pay for underutilized subscriptions. But as Anthropic works toward going public, the company has an incentive to ensure its customer acquisition strategy doesn't lead customers toward rival products or magnify costs.
[12]AI agents promise to 'run the business,' but who is liable if things go wrong?
[13]PrismML debuts energy-sipping 1-bit LLM in bid to free AI from the cloud
[14]Researchers didn't want to glamorize cybercrims. So they roasted them
[15]Netflix - yes Netflix - jumps on the AI bandwagon with video editor
The developer community has long been aware of the price advantage of Claude subscriptions and many subscribers have chosen to use third-party harnesses (e.g. OpenCode, Pi) to interact with Claude Code.
Unfortunately, Anthropic has had trouble keeping up with demand. In February, the biz [16]reaffirmed its preexisting policy forbidding the use of third-party harnesses with Claude subscriptions. This was around the time OpenClaw, an AI agent platform intended to operate autonomously 24/7, began attracting attention.
In late March, Anthropic implemented another strategy to balance demand and capacity: It [17]changed the way subscription usage was calculated so that customers burned through their usage limits faster during peak hours.
As of April 4, 2026, Anthropic went from policy warnings to billing-based enforcement.
"Starting April 4, third-party tools will draw from extra usage instead of subscription limits," an Anthropic spokesperson said in a statement provided to The Register . "Using Claude subscriptions with third-party tools isn't permitted under our Terms of Service, and they put an outsized strain on our systems. Capacity is something we manage thoughtfully, and we need to prioritize customers using our core products."
Claude subscriptions continue to apply to Claude.ai, Claude Code, and Cowork.
Anthropic attempted to mitigate the ill will generated by the move – announced only one day before implementation – by offering subscribers a month of extra usage credit based on their monthly plan. The company is also offering extra usage bundles at 30 percent off. And if that's unsatisfactory, customers have the option to cancel their plan and receive a refund.
Customers can still use Claude with third-party tools through extra usage bundles (purchased through the Claude account page) or by using an API key.
"We've been working hard to meet the increase in demand for Claude, and our subscriptions weren't built for the usage patterns of these third-party tools," Cherny [18]explained . "Capacity is a resource we manage thoughtfully and we are prioritizing our customers using our products and API."
Capacity is a resource that may continue to be scarce if demand continues to grow. Bloomberg recently [19]reported that more than half of the US datacenters planned to open this year will face delays. ®
Get our [20]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2adQtAAlsAZKXELAyKjOk7wAAABA&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44adQtAAlsAZKXELAyKjOk7wAAABA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33adQtAAlsAZKXELAyKjOk7wAAABA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[4] https://status.claude.com/
[5] https://x.com/bcherny/status/2040206440556826908?s=20
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44adQtAAlsAZKXELAyKjOk7wAAABA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://x.com/bcherny/status/2040213608064491525?s=20
[8] https://github.com/google-gemini/gemini-cli/discussions/20632
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33adQtAAlsAZKXELAyKjOk7wAAABA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[10] https://news.ycombinator.com/item?id=44611097
[11] https://she-llac.com/claude-limits
[12] https://www.theregister.com/2026/04/05/ai_agents_liability/
[13] https://www.theregister.com/2026/04/04/prismml_1bit_llm/
[14] https://www.theregister.com/2026/04/05/trellix_john_fokker_roasting_crims/
[15] https://www.theregister.com/2026/04/03/netflix_video_ai/
[16] https://www.theregister.com/2026/02/20/anthropic_clarifies_ban_third_party_claude_access/
[17] https://www.theregister.com/2026/03/26/anthropic_tweaks_usage_limits/
[18] https://x.com/bcherny/status/2040206441756471399?s=20
[19] https://www.bloomberg.com/news/newsletters/2026-04-01/us-data-center-boom-relies-on-hard-to-find-electrical-equipment
[20] https://whitepapers.theregister.com/
"Walled garden" TOS incoming?
Sounds a bit like company-approved AI-app-stores/tool stores being on their way.
I fully expect the AI operators to try to push an increasing amount of lock-in and value-extraction strategies onto their users in the next months.
Somebody really needs to pay for that crazy amount of debt, after all.
Will be interesting to observe the effects on users that already depend on AI.
What if we write our own "claw" tools?
That aint "third-party".
The basic premise is hardly rocket science: loop forever, scraping whatever input stream you choose (e.g. emails, WhatsApp or watching for file updates on your PC); send the scrapings to Claude (or whichever), with a prefix & suffix prompt (which are just strings from config); send the resulting output tokens into bash so it can act out your demands. Maybe add a mode or flag that feeds the bash output back into the input stream for extra joy.
Elevated errors?
You don't say?