In just a few years, Artificial Intelligence (AI) and Software as a Service (SaaS) models have shifted from technology solutions for some pioneering businesses to the operational infrastructure of many economic sectors. Businesses are no longer just buying software, but are hiring data processing capacity, decision-making capabilities, automation services, and AI models that are constantly updated through the cloud.
In this context, many international businesses are prioritizing the choice of commercial arbitration instead of courts to resolve AI and SaaS disputes
This change also entails a completely different type of dispute compared to traditional software contracts. When an ERP system fails, businesses can relatively clearly identify programming or implementation errors. But when an AI system makes the wrong recommendation, the machine learning model changes the results after each update, or the vendor unilaterally changes the algorithm, causing the quality of service to deteriorate, it becomes much more difficult to determine who is responsible.
In this context, many international businesses are prioritizing the choice of commercial arbitration instead of courts to resolve AI and SaaS disputes. However, a big question for Vietnam is whether the current legal framework and commercial arbitration system are really ready to handle these high-tech disputes.
How are AI and SaaS disputes different from regular technology contract disputes?
The peculiarity of AI and SaaS lies in the fact that the object of the contract is no longer a fixed product. In a traditional software purchase contract, the product is handed over almost completely at the time of acceptance. After that, only maintenance or bug fixing remains. In contrast, in the SaaS model, the service is always on the move. The vendor is constantly updating new versions, changing features, patching security vulnerabilities or improving algorithms that customers often have no control over.
For AI, the level of complexity is even higher. Many AI models act as a "black box". Even the developer cannot always fully explain why the system gives a particular result. When damage occurs, determining the fault lies in the input data, the AI model, the training process, or the way the user operates the system is not a simple legal problem.
International practice shows that AI and SaaS disputes often focus on five groups of issues.
The first is disputes over Service Level Agreements (SLAs), especially for uptime, processing speed, or the ability to recover from incidents.
The second is a dispute related to data ownership, training data, and AI-generated output data.
The third is a dispute over intellectual property rights to AI models, source code, algorithms, or AI-generated content.
The fourth is disputes arising from violations of regulations on personal data protection, cyber security or cross-border data transfers.
Finally, there are disputes over liability when AI makes wrong decisions that cause damage to customers or third parties.
These are all types of disputes that traditional contract law has never faced at the current level.
Why is refereeing becoming the preferred choice?
If we look at international trends, it can be seen that the majority of cross-border AI and SaaS contracts stipulate arbitration as a dispute resolution mechanism.
This comes from the peculiarities of the technology sector.
Firstly, AI disputes often contain large amounts of trade secrets, algorithms, source code, and business data. A public trial in court can increase the risk of exposure of sensitive information. In contrast, commercial arbitration has a prominent advantage in confidentiality. Case records, hearings, and rulings are often not made public, helping businesses protect intellectual property and trade secrets.
Second, AI is a field that requires interdisciplinary knowledge between technology, data and law. Judges do not always have deep enough expertise to evaluate machine learning algorithms, the operating mechanisms of large language models or cloud computing architectures. Meanwhile, parties can choose arbitrators with experience in the fields of information technology, intellectual property, data or digital transformation, thereby improving the quality of dispute resolution.
Third, most AI contracts are international.
The provider can set up servers in Singapore, develop the model in the United States, operate through a data center in Europe, but the customer operates in Vietnam. In this context, the flexibility of location, language, applicable law, and the ability to recognize and enforce awards accordingly make arbitration a much more suitable option than court proceedings.
Vietnamese law has a foundation, but there are no specific regulations for AI
From the perspective of current law, Vietnam has not developed a separate law regulating AI or SaaS contracts. However, this does not mean that disputes do not have a legal basis to resolve.
The 2010 Law on Commercial Arbitration allows the settlement by arbitration of disputes arising from commercial activities or disputes that are allowed by law to be settled by arbitration. Meanwhile, the 2015 Civil Code and the 2005 Commercial Law are still the foundations for regulating the conclusion, performance and breach of contracts.
From a data perspective, the 2024 Data Law, the 2025 Personal Data Protection Law, Decree No. 356/2025/ND-CP and the 2025 Cybersecurity Law have formed a new layer of legal obligations for businesses that develop or use AI.
This means that if an AI system breaches its obligations to protect personal data or processes data illegally, contractual disputes will not only be a breach of civil obligations, but may also entail administrative liability or other legal obligations.
It can be said that Vietnam has a "background legal framework", but there are no specialized regulations reflecting the unique characteristics of AI. That is the biggest gap between traditional law and current technological practices.
The biggest gap is not in arbitration law
Many argue that the Law on Commercial Arbitration is outdated before the development of AI.
In fact, the problem does not lie entirely in this law itself. The Arbitration Law mainly regulates dispute resolution procedures. Meanwhile, the biggest difficulty lies in the content of the dispute.
For example, when AI automatically updates the model and produces results that are different from when the contract was signed, is this the fault of the supplier? If the AI makes the wrong decision but the supplier has warned in advance about the limitations of the model, how will the responsibility be divided?
Or when AI is trained on customer data, who will have ownership of the newly formed model belong? These are all questions that current contract law has not directly answered.
That makes arbitrators forced to rely more on contractual terms, international commercial practices, the principle of goodwill and the will of the parties.
In other words, in AI disputes, the quality of the contract often determines the outcome of the dispute more than the law itself.
It is an obvious fact that the speed of AI development is always faster than the speed of amending laws.
Electronic evidence will become at the center of every AI lawsuit
A huge difference between AI disputes and traditional commercial disputes lies in evidence. In many cases, evidence is no longer emails or paper contracts.
Evidence can be system logs, training data, algorithm update history, API logs, metadata, cloud-based logs, or user and AI interaction history. If these data are not stored properly, it is almost impossible for businesses to prove the other's faults.
This poses new requirements for both businesses and arbitration centers. Enterprises need to develop a process for storing electronic evidence from the beginning instead of only collecting them when a dispute has arisen. Meanwhile, arbitrators need to understand how to assess the integrity, origin and evidentiary value of electronic data, rather than focusing only on traditional evidence.
In the future, digital experts, AI experts, and data forensics experts are likely to become just as important as lawyers in technology arbitration cases.
What businesses need to do is not wait for the law to be completed
It is an obvious fact that the speed of AI development is always faster than the speed of amending laws.
This is not only happening in Vietnam but also in most countries. The European Union has enacted the Artificial Intelligence Act in 2024, and many international arbitration centers such as SIAC, HKIAC, ICC or WIPO are also handling more and more disputes related to AI, data and digital technology. However, even in countries with developed legal systems, most issues are still solved through contract design and risk management, not by detailed legal provisions.
For Vietnamese businesses, the most important lesson is not to wait for a separate law on AI. What is more essential is to build AI and SaaS contracts with a higher level of detail than before. Provisions on data rights, the right to use AI outputs, service quality standards, responsibility for AI models, algorithm update processes, audit mechanisms, information security, personal data protection, and especially dispute resolution clauses need to be designed right from the negotiation stage.
Conclusion
AI and SaaS are changing the way businesses enter, execute, and exploit contracts. At the same time, they are also posing entirely new challenges to dispute resolution mechanisms. Vietnamese law now has a sufficient foundation to handle many issues through contract, commercial, arbitration and personal data protection law. However, that platform is only a necessary condition.
Sufficient conditions will come from the simultaneous development of three factors: more efficiently designed contracts, a team of arbitrators with a deep understanding of the technology, and the gradual perfection of the legal framework in accordance with the characteristics of AI. As technology evolves faster than law, the adaptability of businesses and of dispute resolution mechanisms will determine whether commercial arbitration will continue to be an effective body for AI and SaaS disputes in the future.
