OpenAI vs Anthropic Clash 2026: User Banned for GPT on Claude Code
OpenAI and Anthropic execs clashed publicly after a user was banned for running GPT on Claude Code. Here is what happened and why it matters to you.
📰 What Actually Happened
A user posted online that Anthropic had suspended their account after they ran GPT-5.6-Sol, an OpenAI model, inside Claude Code, which is Anthropic's popular AI coding tool. The complaint reached Thibault Sottiaux, who leads OpenAI's Codex product, and he replied with a jab: "I'd love to help, but I don't work at Anthropic," adding that OpenAI is hiring.
Boris Cherny, who leads Claude Code at Anthropic, pushed back quickly. He said Anthropic does not ban people for using its coding harness with other models, and that the suspension was almost certainly triggered by a separate account classifier. He said he was looking into the case personally.
Sottiaux then escalated the friendly fire. In a follow-up, he wrote "I'm not going anywhere" and announced that OpenAI had reset usage limits for all paid users of ChatGPT Work and Codex. That move turned a single user complaint into a public marketing skirmish between two of the biggest AI labs in the world.
What is Claude Code, in plain English?
Claude Code is Anthropic's AI agent that lives in your terminal or desktop and does real work for you: writing code, editing files, and running tasks. Think of it as the 'body' (the harness) and the AI model as the 'brain.' Some technically minded users swap in a different brain, like an OpenAI GPT model, while keeping the Claude Code body. That mix-and-match setup is exactly what this story is about.
⏪ The Backstory: This Rivalry Did Not Start Yesterday
This exchange lands on top of an existing feud. Back in 2025, Anthropic cut off OpenAI's access to the Claude API, accusing OpenAI staff of using Claude to benchmark and test GPT-5 in violation of its terms of service. OpenAI argued that benchmarking rival models is standard industry practice. Anthropic held its position, and the relationship has been frosty since.
Both companies now compete head-to-head in AI coding tools. Anthropic ships Claude Code, and OpenAI ships Codex. Their product leaders, Cherny and Sottiaux, are visible on social media and clearly enjoy the rivalry. Sottiaux has even reset Codex usage limits before as a competitive flex.
For readers, the important nuance is this: the user's ban was reportedly not a policy of punishing people for using rival models. Cherny stated the opposite. But the incident shows how quickly a support ticket can become ammunition in a platform war.
💡 Why This Matters for Solopreneurs and Everyday Users
First, it confirms a trend that benefits you: AI tools are becoming mixable. The harness (the app that does the work) and the model (the intelligence) are separating. You can increasingly pick the best model for each job, the way you pick apps on a phone. When executives publicly say 'we do not ban users for using other models,' that statement becomes a public commitment users can point to later.
Second, it is a reminder that your account can get flagged by automated systems even when you did nothing against the rules. Cherny blamed an account classifier, which is an automated safety filter, not a human decision. If your business depends on one AI subscription, an automated false positive can freeze your workflow for days. That is a real operational risk for a one-person business.
Third, competition is producing free wins. OpenAI responded to the drama by resetting usage limits for paid ChatGPT Work and Codex users. When two well-funded rivals fight for the same developers and knowledge workers, prices, limits, and features tend to move in the user's favor. Watching these clashes is not just entertainment, it is market intelligence.
⚖️ Claude Code vs OpenAI Codex: Quick Comparison
If the headline made you curious about the two products at the center of this clash, here is a plain-language comparison. Both are AI coding agents: you describe what you want, and they do the work. The differences that matter for non-developers are the ecosystem, the default models, and how open each tool is to outside models.
| Aspect | Claude Code (Anthropic) | Codex (OpenAI) |
|---|---|---|
| Made by | Anthropic | OpenAI |
| Default brains | Claude models (e.g. Claude Sonnet 4.6 and newer) | GPT models (e.g. the GPT-5.x line) |
| Where it runs | Terminal, desktop app, web, IDE extensions | Terminal (CLI), IDE, and cloud tasks |
| Third-party models | Cherny says users are not banned for routing other models through it | Open-source CLI, community adapters exist |
| Best fit today | Users already paying for Claude | Users already paying for ChatGPT |
🛠️ How to Act on This Today
You do not need to be a developer to take something practical from this story. The core lesson is resilience: do not let a single AI account become a single point of failure for your work.
If you want to try an AI coding agent, start with the official setups. Install Claude Code from Anthropic or Codex from OpenAI, use it with its native models first, and learn what it can automate in your daily work: drafting scripts, cleaning spreadsheets, or building small tools. Mixing models across tools is possible through community routers, but read each provider's terms of service first and know that automated flags can happen.
If you are a paid ChatGPT Work or Codex user, check your usage dashboard. The reset Sottiaux announced means you may have a fresh allowance right now, which is a good moment to run a bigger experiment you had postponed.
- ✔Keep at least two AI subscriptions or free-tier fallbacks so one ban or outage cannot stop your work
- ✔Export or back up important chats, prompts, and generated files weekly
- ✔Read the terms of service before routing one company's model through another company's tool
- ✔If you get flagged unfairly, appeal through official support first, then post politely and publicly with details
- ✔Check your ChatGPT Work or Codex usage limits now to see if the reset applies to you
🔭 What to Watch Next
Watch whether the banned user's account gets restored, since Cherny said he was looking into it. A quick, public fix would back up Anthropic's claim that this was a classifier error, not a policy.
Also watch whether either company formalizes its stance on third-party models in writing. Public statements from executives on social media are encouraging, but terms-of-service language is what actually protects users. The 2025 dispute, where Anthropic revoked OpenAI's Claude API access over GPT-5 benchmarking, shows both labs will enforce their terms aggressively against each other.
Finally, expect more perks. Usage limit resets, bundled features, and aggressive pricing are the visible edge of this rivalry, and users of both platforms stand to gain through the rest of 2026. Sources for this story include Inshorts, TechCrunch, and Tom's Guide coverage of the Anthropic and OpenAI dispute.
❓ Frequently Asked Questions
Can you really run GPT models inside Claude Code?
Technically, yes. Claude Code is a harness, and community tools let advanced users route other models, like OpenAI's GPT-5.6-Sol, through it. Boris Cherny, who leads Claude Code at Anthropic, stated that Anthropic does not ban users for doing this. Still, always check the current terms of service for both companies before mixing, since policies can change.
Why was the user's Anthropic account suspended?
According to Cherny's public response, the suspension was almost certainly triggered by an automated account classifier, not by a rule against using rival models. He said he was investigating the case. This means the ban was likely a false positive from a safety system rather than deliberate punishment.
What is the history between OpenAI and Anthropic?
Anthropic was founded in 2021 by former OpenAI employees, and the two labs are direct competitors. Tensions became public in 2025 when Anthropic revoked OpenAI's Claude API access, accusing OpenAI of using Claude to benchmark GPT-5 in violation of its terms. OpenAI called benchmarking standard practice. This latest exec exchange is a continuation of that rivalry.
What does 'I'm not going anywhere' mean in this story?
The quote comes from OpenAI Codex chief Thibault Sottiaux during the public exchange. He paired it with an announcement that OpenAI had reset usage limits for all paid ChatGPT Work and Codex users, signaling that he and his product intend to keep competing hard for coding users.
🏁 Final Thoughts
The short version: a user got flagged after running OpenAI's GPT-5.6-Sol inside Anthropic's Claude Code, two high-profile execs traded jabs, Anthropic said the ban was a classifier error and not policy, and OpenAI turned the moment into a usage-limit giveaway. For solopreneurs and knowledge workers, the takeaways are practical. AI tools and models are becoming mixable, automated bans can hit anyone, and fierce competition keeps delivering free perks. Build redundancy into your AI stack today so no single account failure can stop your business. If this explainer saved you a research rabbit hole, subscribe to Agents at Work for plain-English AI news breakdowns, and drop a comment telling us which AI coding tool you use and why.
Last updated: August 09, 2026 · Keyword: OpenAI Anthropic clash Claude Code · Agents at Work
Comments
Post a Comment