OpenAI Admits GPT-5 Misstep Amid User Backlash and Shifts Focus to GPT-6

OpenAI CEO Sam Altman has acknowledged that the launch of GPT-5 was a misstep, admitting the company “screwed up” after widespread user backlash over the replacement of the popular GPT-4o model [1]. Users quickly criticized the new version, describing it as colder, more mechanical, and less supportive. In response, OpenAI restored access to GPT-4o for paying subscribers, a reversal Altman called a “wake-up call” that highlighted the importance of user experience when rolling out upgrades to products used by hundreds of millions [1].

During a private dinner in San Francisco, Altman reflected on the emotional connection users had formed with the AI model, particularly with GPT-4o, which some described as a “companion.” While he estimated that fewer than 1% of ChatGPT users had “unhealthy relationships” with the bot, he stressed the need to remain vigilant about how AI systems are designed to interact with people [1].

Looking ahead, Altman is now directing attention toward GPT-6. He emphasized that the next iteration must maintain a sense of personality without potentially exploiting users. However, he noted that the rollout of GPT-6 would face significant challenges, particularly due to a shortage of GPUs. Altman stated that OpenAI would need to invest “trillions of dollars on data center construction in the not very distant future” to overcome these bottlenecks [1].

Despite the rocky start with GPT-5, ChatGPT’s user base continues to grow, with the platform now reaching over 700 million weekly users. Altman, however, expressed concerns about the broader AI industry, warning that an AI bubble may be forming. He acknowledged that investors are currently overexcited about AI, but also affirmed that AI remains one of the most transformative developments in recent history [1].

In addition to refining its language models, OpenAI is pursuing a broader strategic vision. Altman outlined initiatives such as supporting a brain-computer interface startup and collaborating with Jony Ive, Apple’s former design chief, on a yet-unannounced AI device. These efforts reflect OpenAI’s ambitions to expand beyond conversational AI and into new technological frontiers [1].

Sources:

[1] OpenAI CEO Sam Altman Concedes GPT-5 Was a Misfire ... (https://decrypt.co/336053/openai-ceo-sam-altman-concedes-gpt-5-misfire-bets-gpt-6)