In a move that sparked widespread debate, OpenAI CEO Sam Altman announced the reasons behind the company’s shift toward closed artificial intelligence models, a transition that has caught the attention of industry experts and observers. During an “Ask Me Anything” session hosted by Altman on Reddit, he explained that this change aims to achieve higher security standards, helping ensure the safety of the AI applications provided by the company.
The reason for shifting to closed AI models
OpenAI began its journey with principles of transparency and public participation, launching its models as open-source projects. This approach was part of its strategy for innovation and attracting top talent. However, in recent years, the company has begun moving toward a more conservative and closed approach. Altman explained that this shift resulted from the company’s evaluation of the security levels achievable through more controlled, closed models.
Responding to a Reddit user’s question about why OpenAI’s models are not made open source like ChatGPT, Altman stated that the company concluded that using closed models makes it easier to achieve desired safety standards. He noted that providing well-considered APIs and services can be safer than making every model publicly available, which helps protect users and prevents the misuse of advanced technology.
He also indicated that the company looks forward to opening up some of its technologies in the future, explaining that the role of open-source technology remains important in the AI ecosystem.
The controversy over OpenAI’s shift and abandoning transparency
This shift toward closed models at OpenAI is not merely an internal decision; it has provoked strong reactions from prominent figures in the tech sector, including Elon Musk. Musk, who co-founded OpenAI in 2015 and later withdrew from the board of directors, has not stopped criticizing the company’s trajectory.
In December 2022, Musk tweeted pointing out that OpenAI started as a non-profit, open-source organization, but emphasized that the company is no longer structured that way. Musk filed a lawsuit against the company in March of this year, accusing it of straying from its original purpose. Although he dropped the lawsuit in June, he filed a new one last August, claiming he was deceived into supporting the company at its founding.
These escalating disputes have ignited discussions about the nature of major tech companies and their role in offering artificial intelligence technologies as public open projects. This ongoing controversy brings questions back to the forefront regarding the risks and ethics of making advanced AI technologies available to the public.
Meta’s stance on open artificial intelligence
In this context, Meta serves as an example of a company following an open approach to AI development, recently launching the Llama 3.2 model as an open-source project last September. CEO Mark Zuckerberg stated that openness in AI development is safer for ensuring a positive future for technology, emphasizing that openness increases safety levels compared to closed options.
Nevertheless, the Llama model is not available to everyone unconditionally; companies with more than 70 million monthly active users are required to apply for permission to use it, leading to debate over the extent to which Llama complies with open-source standards.
In this regard, the Open Source Initiative (OSI) issued a statement noting that the Llama license does not align with the open-source definition because the model contains certain restrictions. Meta responded that it disagrees with the OSI’s definition, explaining that providing Llama free and accessible to everyone remains consistent with openness goals, even if certain restrictions exist to maintain security.
What’s next for OpenAI and the future of AI?
Returning to OpenAI, Altman hinted during the session that the company might return to openness in certain areas in the future, expressing a desire to open more technology to the public once an advanced level of safety is reached. He added that the company strives to make AI safely and sustainably available to users, especially amid growing challenges related to security and malicious applications.
He also indicated that the launch of GPT-5 will not happen this year, as the company is keen to develop it slowly to ensure compliance with high safety standards. This came in the context of Altman using the ChatGPT app while answering questions during the Reddit session.
The future of AI between openness and restriction
The AI sector today is witnessing increasing competition between companies striving to strike a balance between openness and security, especially as technology evolves and its impact on daily life magnifies. While some, like Meta, adopt an open approach, OpenAI prefers to control its models, believing this approach achieves higher security levels.
Ultimately, the decision to open or close AI development remains a vital topic, as each direction carries implications for the industry and the future of technology, forcing companies to adapt to security challenges, malicious applications, and growing user demands.
