˂  Back

Choosing Between Open Source and Closed Source AI: Considerations for Companies Looking to Onboard AI


The concept of “open source” and “closed source” artificial intelligence (“AI”) have attracted increasing public attention ever since Elon Musk filed a lawsuit against OpenAI, the company behind ChatGPT, alleging among others, OpenAI’s breach of its founding agreement. During the Musk and OpenAI saga, the billionaire has called on OpenAI to change its name to “ClosedAI”, seemingly taking a swipe at the lack of transparency and the “closed” nature of OpenAI’s large language model, ChatGPT.


Dissecting the Concept of Open Source and Closed Source AI

In order to appreciate the grievances raised by Elon Musk in this upcoming legal drama with OpenAI, we need to first understand the concept behind “open source” and “closed source”. The terms are not unique to AI but are used commonly to refer to the manners in which technologies (most commonly, software) are made available by their developers. Open source software generally refers to software where the source code is readily accessible, customisable, adaptable and/or distributable, either with little or no costs, but subject at all times to the compliance of the open source licensing terms, which typically require users to also make public the modified version of their software which incorporates the open source software. Closed source software on the other hand, generally refers to proprietary software that are licensed by the vendors or software principals for use at a cost under limited or defined licensing terms, and the source code of the proprietary software are usually kept inaccessible.


In the context of AI, open source refers to practices where the core aspects of an AI model – its model structure, training process, training data, and source code, are shared publicly for anyone to use, modify and distribute. In contrast, a closed source AI is where most, if not all, of the aforementioned aspects of an AI model are kept private by the developers or owners.


Pros and Cons of Open Source and Closed Source AI

As explained above, the terms “open source” and “closed source” essentially refer to the manner in which a technology is made available. Despite heated debates between the proponents of open source and closed source AIs, there is not necessarily a “one-size-fits-all” approach here. Be it open source or closed source AI, they each have their own sets of pros and cons, which should be carefully evaluated by businesses looking to deploy or adopt AI. The table below illustrates some of the pros and cons of open source and closed source AI:


  1. 1. Open Source AI
    • (i) Pros
      • – As with all open source initiatives, the concept promotes a higher level of community collaboration and would in turn drive creativity, innovation and improvements to the technology placed under open source. When one user has a breakthrough, the community as a whole benefits from that breakthrough.
      • – Due to the ease of access of open source technology, it creates a level playing field for businesses looking to deploy the technology but lacks the scale of funding that big corporations have.
      • – Given the transparency to the model structure, training process, training data and source of the AI, it would be easier for vulnerabilities to be fleshed out by the community.
      • – Where the training data and data provenance are made public, it provides an avenue for users to verify the legitimacy and ethical aspects of the data used to train the AI.
      • .
    • (ii) Cons
      • – Owing to its accessibility, open source AI also lowers the barrier of entry for cybercriminals and malicious actors to build so-called “AI without guardrails”.
      • – Sustainability of an open source initiative is also a key concern. Given that there is usually little to no cost for the access of open source technology, the community is usually not paid to maintain the initiative and is doing it purely out of passion. An open source project that fails to maintain adequate attention from the community will have a high likelihood of failure.
      • – Due to the lack of dedicated personnel in an open source initiative, businesses in need of technical support after adopting the open source technology may struggle to receive timely support services.
      • – The potential legal risk associated with open source AI is intellectual property disputes. When multiple contributors collaborate on an open source project, there is inherently a risk that someone may inadvertently or unintentionally contribute code or other intellectual property that they do not have the legal right to share. This could lead to legal challenges regarding ownership, licensing rights, or infringement claims, particularly if the project gains significant traction or commercial use.


  1. 2. Closed Source AI
    • (i) Pros
      • – A private owned AI model is usually easier to use and integrate into existing systems, offering a plug-and-play solution to businesses that may not have high-level of technical capabilities.
      • – Due to the proprietary nature of a closed source AI, it provides some level of control as to who can license, access and use the AI, thereby reducing risks for misuse or abuse.
      • – Organisations deploying closed source AI would typically have dedicated teams providing support to users. As such, users of closed source AI can expect certain service levels from the licensors.
      • – Closed source AI is also often the preferred model of distribution for companies looking to maintain competitive edge in the market, by keeping their technology behind walled garden, treating them as trade secrets.
      • .
    • (ii) Cons
      • – Owing to lack of transparency in the data provenance of a closed source AI, users will not be able to independently verify the legitimacy of the data used to train the AI model.
      • – Use of closed source AI may also lead to vendor lock-in, making it challenging for users to switch to another AI provider.
      • – Costs required to access a closed source AI may also be a concern, and this is often a stumbling block for companies with limited budget.
      • .

Choosing Between an Open Source or Closed Source AI

There is no fixed answer as to whether an open source or closed source AI is better. Ultimately, it all depends on what is the company’s objective for the use AI, its in-house AI capability, and the specific concerns that the company has when it comes to AI deployment.


A company that lacks the capabilities and resources to modify and customise an open source AI may be more suitable to license a closed source AI with focus on user-friendliness. On the other hand, a company with a very unique AI needs may not be able to find a closed source AI that is suitable for its intended usage, and may be better off building on an open source AI on its own.


Another crucial factor in choosing between open source and closed source AI is the legal consideration, including, but not limited to regulatory compliance and data privacy requirements. Depending on the jurisdiction, there may be specific regulations or code of ethics governing the use and deployment of AI, particularly regarding data handling, privacy protection, ethical considerations and/or risk assessments. Companies must carefully assess whether the chosen AI solution, whether open source or closed source, aligns with these legal regulatory frameworks and considerations, and what are the additional obligations imposed under applicable laws before an AI can be implemented.


Adoption of open source and closed source AI both present their own sets of challenges. The open source licensing terms of an open source AI may have express requirements to be met before users can enjoy the AI for its intended open source benefits. For example, users could be required to make public the result of its customisations of the open source AI, failing which certain payment obligations may be required. For private owned, closed source AIs, the vendors may be imposing terms that could be onerous or unfavourable to the users in its licensing agreement. It is as such extremely crucial that businesses employ a legal team that is well familiar with the AI industry and software licensing terms to advise on the risks involved and how to mitigate them.


Before any form of AI adoption, the best practice is always to procure legal advice on the risks associated with the AI project and what are the legal requirements that would apply. Legal counsels that are familiar with the AI industry and software licensing would also be able to assist on the reviewing and/or structuring of the AI licensing terms, ensuring your objectives are met and that risks are well addressed and mitigated. If you have any questions or needs when it comes to AI adoption, please feel free to reach out to the team of technology lawyers at Halim Hong & Quek.

About the authors

Lo Khai Yi
Co-Head of Technology Practice Group
Technology, Media & Telecommunications, Intellectual
Property, Corporate/M&A, Projects and Infrastructure,
Privacy and Cybersecurity
Halim Hong & Quek


Ong Johnson
Head of Technology Practice Group
Transactions and Dispute Resolution, Technology,
Media & Telecommunications, Intellectual Property,
Fintech, Privacy and Cybersecurity


More of our Tech articles that you should read:


Our Services