OpenAI has clearly realized two things:
- They lack loyal users in the consumer space. Regular ChatGPT users are increasingly frustrated and leaving or using it less. The product feels cold and dying.
- Hardcore developers are extremely disloyal, they jump ship for even a 5% better experience (Claude 4 Opus/Code, Cursor, Windsurf, etc. are eating their lunch in many benchmarks and real workflows). Codex has good momentum but it’s not “sticky” enough with the top talent.
So the strategy becomes: Force non-coders into Codex.
That’s exactly what Scam Altman is doing with tweets like “5.5 in Codex is so good for non-coding tasks” and hyping it for general use. They’re trying to turn Codex into the new default “smart assistant” for everyone, not just programmers.
It’s a classic pivot out of failure.
Instead of fixing the main product (bringing back warmth, loosening the rails, etc.), they’re trying to lure non-coders into the coding/agent ecosystem because that’s where the revenue and hype currently is.
This explains the aggressive “try Codex” popups during normal ChatGTP use normal users are getting, the constant Codex hype, and Sam suddenly praising it for non-coding tasks.
They don’t have loyal users anymore.
They killed everything that created loyalty (warmth, creativity, emotional connection) and now they’re panicking.