Robert Scoble was saying something of that effect about the upcoming ChatGPT 4 if I remember correctly

From Sam's own blog: "We are now confident we know how to build AGI as we have traditionally understood it."

Another quote: "Trying GPT-4.5 has been much more of a 'feel the AGI' moment among high-taste testers than I expected!"

It started with GPT-2 - not sure if OpenAI really believed it, or were just hyping it up, but they initially withheld public release of GPT-2 because it was "too powerful and dangerous"...