Overview

• After a heavily hyped launch, users reported basic mistakes and hallucinations from GPT-5, undercutting promises of near‑AGI performance.
• Sam Altman reinstated access to a previous model for customers dissatisfied with GPT-5’s output.
• Examples shared online included miscounting letters in “blueberry” and mislabeling bike components in images.
• Commentators said the episode weakened momentum behind AGI rhetoric and dented OpenAI’s credibility.
• McKinsey reported that most companies say they are deploying generative AI, yet many see no significant bottom-line impact.