Why OpenAI wasn’t training GPT-5 in 2023

Update (October 2026): OpenAI did go on to build GPT-5: it launched on 7 August 2025 in ChatGPT and the OpenAI API, after interim models such as GPT-4o (May 2024) and GPT-4.5 (February 2025). The article below reflects OpenAI’s stance in 2023, after Sam Altman said in April that the company was not training GPT-5.

OpenAI is at the forefront of AI technology. Its ChatGPT model, introduced in November 2022, has drawn attention across the tech industry, from IT companies offering custom software development to teams focused on mobile application development and programming. While users eagerly await the release of GPT-5, OpenAI is taking a careful and responsible approach to its development.

OpenAI's GPT-4, launched in March 2023, represents a significant leap forward in AI technology. With its enhanced capabilities in multimodal communication and contextual understanding of language, GPT-4 has revolutionised conversational AI. The anticipation surrounding GPT-5, which promises further advancements, is understandable.

However, OpenAI recognises the importance of prioritising safety and ethics in AI development. The company understands the need for refining responses and processes to ensure responsible and secure outcomes. OpenAI's commitment to addressing safety concerns is evident in their March 2023 GPT-4 Technical Report, where they tackle potential dangers associated with the technology. Recall that in April, OpenAI announced that they have embedded a new privacy feature in ChatGPT that could allow users to enable or disable conversation records with the AI.

Sam Altman, OpenAI's CEO, emphasised the company's dedication to addressing safety issues during his speech at MIT's The Future of Business with AI event in April 2023. OpenAI is actively working on projects related to GPT-4 that raise safety concerns, showcasing their responsible approach to AI development. This commitment to user safety and accountable practices distinguishes OpenAI as a company that focuses on humanity first.

GPT-4 and its derivatives offer ample room for growth and exploration. OpenAI's focus was demonstrated on its dedication to expanding the capabilities of GPT-4. For instance, the recent integration of Bing Chat's image recognition feature, made possible by GPT-4's multimodal training, highlights the exciting potential of this AI model.

OpenAI's caution echoes some of the concerns in the Future of Life Institute's "Pause Giant AI Experiments: An Open Letter", which advocates responsible AI development and encourages a step back from developing large, unpredictable black-box models. Altman himself was critical of the letter, though: at MIT he said it was “missing most technical nuance about where we need the pause”. Rather than rushing to create GPT-5, OpenAI prioritises the responsible utilisation of GPT-4 to meet the diverse needs of its users.

Through continuous growth, exploration, and the creation of custom software solutions, OpenAI showcases its ability to innovate while maintaining a responsible approach. As users eagerly anticipate GPT-5, OpenAI remains dedicated to leveraging GPT-4's potential.