OpenAI product lead Tibo recently showed users how to give Claude Code a brain transplant: keep Claude Code's harness, swap the underlying model for GPT.
Developer Alex Getman followed the setup — and his Claude account was banned pretty quickly. He publicly appealed to Tibo and Anthropic, asking whether this kind of setup was actually against the rules.
Tibo then got passive-aggressive: "I'd love to help, but unfortunately I don't work at Anthropic." He added that getting banned for using Claude Code's harness with a different model was "indeed pretty weird," and asked whether anyone else had been caught up in it.
Claude Code lead Boris Cherny jumped in quickly. He made clear Anthropic doesn't ban people for using the harness with other models, and that this was almost certainly some other risk control misfiring. The account was restored shortly after.
Boris also took a shot back at Tibo: "If you want to come to Anthropic, we're hiring." Tibo said he loves his current team and has plenty of new products to ship in the coming weeks, so he's not going anywhere.
Once everything was resolved, Tibo posted a celebration, noting GPT-5.6 Sol works almost everywhere — including the Claude Code harness. To mark the occasion, and the fact that he's not jumping ship, he reset usage limits for all paid ChatGPT Work and Codex users.
https://twitter.com/thsottiaux/status/2086188036493344823
