Over 1,200 likes on a single X post. That’s how fast Peter Steinberger’s beef with Anthropic blew up Friday morning.
He shared a screenshot: Anthropic had suspended his Claude access for ‘suspicious activity.’ Irony? Steinberger — now at OpenAI — built OpenClaw, the open-source harness that’s been letting devs run agents across AI models, including Claude. And folks, this hit right after Anthropic slapped a paywall on tools like his.
Look. Subscriptions to Claude used to cover third-party harnesses. No more. Now OpenClaw users pony up via API, consumption-based. Anthropic calls it a ‘claw tax’ fix for heavy usage — loops, retries, tool chains that chew compute.
But Steinberger wasn’t having it. ‘Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source.’
What Triggered the Ban on OpenClaw’s Creator?
Steinberger swears he was playing by the new API rules. Testing OpenClaw updates to keep it humming on Claude — for users who pick it over ChatGPT. (Yeah, Claude’s still king for some OpenClaw fans.)
Suspension hits. Post goes viral. Hundreds of comments, conspiracy vibes thick — guy’s at OpenAI now, after all. Then an Anthropic engineer chimes in: ‘We’ve never banned anyone for OpenClaw. Happy to help.’
Account back online hours later. Coincidence? Maybe. But it reeks of PR whack-a-mole.
“Yeah folks, it’s gonna be harder in the future to ensure OpenClaw still works with Anthropic models,” OpenClaw creator Peter Steinberger posted on X early Friday morning, along with a photo of a message from Anthropic saying his account had been suspended over “suspicious” activity.
That quote? Straight fire. Captures the chill in the air between open tools and closed gardens.
Here’s my take — the one you won’t find elsewhere. This echoes the Netscape wars of the ’90s. Microsoft crushed the upstart browser by bundling IE, twisting Windows APIs. Anthropic? They’re not there yet, but Cowork — their agent — just got Claude Dispatch. Remote control, task assignment. Sounds familiar? Rolled out weeks before the pricing pivot. OpenClaw features, now proprietary. History rhymes: control the stack, kneecap the rivals.
Why Did Anthropic Slap a ‘Claw Tax’ on Claude Subscriptions?
Subscriptions weren’t built for agent workloads, Anthropic says. Fair point — claws loop reasoning, retry endlessly, hook into APIs galore. Spikes costs.
But timing. Dispatch drops. Pricing changes. Steinberger calls copycat. And he’s not wrong: Cowork’s eating OpenClaw’s lunch for enterprise users who want plug-and-play agents.
Steinberger fires back on X: ‘One welcomed me, one sent legal threats.’ (Anthropic’s past? They’ve lawyered up on devs before.) Ouch. Points to a pattern — protect the moat at all costs.
Users grill him: Why test Claude over OpenAI’s stuff? ‘Separate my OpenClaw Foundation work — make it model-agnostic — from OpenAI strategy gigs.’ Sneaky clue: he’s eyeing OpenClaw tweaks for GPT dominance.
Is Anthropic’s Move a Death Knell for Cross-Model Tools?
Not yet. Ban lifted. Engineer outreach. But the tax sticks. Devs grumble — why pay extra when prompts are cheap?
Anthropic’s betting heavy users (you know, the ones building real agents) will migrate to Cowork. Smart business. Skeptical? It’s corporate Darwinism. OpenAI did similar with custom GPTs; now everyone’s circling wagons.
Bold call: Expect more. As models specialize — Claude for code, GPT for chat — harnesses like OpenClaw become glue. Glue threatens silos. Watch for API rate limits, ‘terms violations,’ anything to steer you in-house.
Steinberger’s no victim. iOS dev turned AI tinkerer, now OpenAI strategist. But his saga spotlights the fragility: one viral post flips the script. Power to the people — for now.
Frustrations boiled over in replies. One jab: ‘You went to the wrong company.’ Steinberger: legal threats from Anthropic. Public memory’s long.
And the engineer? Genuine olive branch or damage control? We’ll see — I pinged Anthropic too. Crickets so far.
This isn’t just drama. It’s architecture shifting underfoot. OpenClaw thrives on universality. Anthropic wants loyalty. Collision course.
Why Does OpenClaw Still Matter for Claude Users?
Popularity persists. Claude edges ChatGPT in agent tasks for some — reasoning depth, maybe. Steinberger’s testing keeps it alive.
But hints at OpenAI pivot: ‘Working on that.’ Translation? Supercharge OpenClaw compatibility there. Rival poach in motion.
Devs, heads up. Multi-model betting? Riskier daily.
**
🧬 Related Insights
- Read more: Jassy’s Shareholder Letter: Amazon’s Chip Assault on Nvidia’s Empire
- Read more: OpenAI’s $852 Billion Bet Faces Exec Exodus: Simo and Rouch Step Back for Health
Frequently Asked Questions**
What is OpenClaw and why was Anthropic’s creator banned?
OpenClaw is an open-source agent harness for running AI tasks across models like Claude. Peter Steinberger got temporarily banned for ‘suspicious activity’ while testing via API post-pricing change.
Will Anthropic block OpenClaw from Claude forever?
No full block — access reinstated, but subscriptions no longer cover it. API pay-per-use now, targeting heavy agent workloads.
Does this mean AI companies hate open-source tools?
Not hate, but competition. Anthropic pushes its Cowork agent; pricing protects margins amid copycat features.
