by Atek Shuvo (Policy Intern at CELI)
The central mechanism of international investment law is the dispute settlement process and it’s a controversial set up.[1] The main agenda is to protect foreign investors from arbitrary and discriminatory state conducts and it allows claims to be brought against host states.[2] Investor State Dispute Settlement (ISDS) mechanism faces a lot of criticism for a lack of transparency, inconsistency, high costs and potential chilling effects on domestic laws.[3] Simultaneously, there is a shift on international legal regime as artificial intelligence (AI) is being drastically deployed in every sector and interfering in the current set up of governance, regulations, dispute resolutions and legal practices.[4] So, the question arises whether the intersection of ISDS and AI can lead to a potential imbalance to resolve the issues related with legitimacy, efficiency, accountability and fairness in international investment laws.
Here, the discussion will revolve around how this intersection may further interfere with ISDS. Also, it will evaluate the role of AI during the ISDS proceedings as well as investors conducts and state regulations.
AI, State Regulation, and Investment Disputes
It is evident that AI is dramatically becoming the essential apparatus of states regulatory frameworks.[5] Nowadays, governments are gradually relying on AI for tax administration, customs valuation, licensing, environmental monitoring, financial supervision, and public procurement.[6] While these technologies promise efficiency and objectivity, they also generate new risks. Algorithmic opacity, data bias, and automated decision-making may produce outcomes that foreign investors perceive as arbitrary, discriminatory, or disproportionate.[7]
But then again if we evaluate the current practices, we can easily assume that there are possibilities where AI can create complications during the assessments of general treaty standards, due process, non-discrimination and fair and equitable treatments.[8] To illustrate, imagine if an AI based licensing system denies an investor approval due to flawed data or biased training sets, questions arise as to whether the state has breached its international obligations. Here, ISDS tribunals need to need to grapple with whether algorithmic decisions can be attributed to the state, how foreseeability and transparency should be assessed in automated systems, and whether reliance on AI alters the standard of reasonableness expected from public authorities.
However, investors can deploy AI in their operations in several sectors including finance, logistics and digital services. And, it may create disputes and these emerging disputes can be related with indirect expropriations or undue interference because AI based business models like data localizations or algorithmic accountability can be regulated by the host states.[9] So, it can be easily coined that AI can reshaped the elements of investment disputes.
AI in ISDS Proceedings
Currently AI started to influence the conducts of ISDS proceedings. Legal practitioners already use AI-powered tools for document review, legal research, and case prediction.[10] The ISDS cases where thousands of papers and documents are involved, AI assisted documents management system can reduce time and cost dramatically.
In addition, AI has ability to affect the decision making itself. Predictive analytics can identify patterns in arbitral awards, helping parties assess litigation risk and settlement strategies.[11] These tools have the ability to enhance efficacy but also have the risk of reinforcing biases or can encourage the strategic behavior that sometimes prioritizes predictability over legal reasoning.
The potential use of AI generated or AI assisted arbitral awards can raise even fears. While fully automated arbitration remains unlikely in the near future, partial reliance on AI for drafting or reasoning could challenge core principles of ISDS, including arbitrator independence, due process, and the right to be heard.[12] There can be possibilities where the parties of the disputes can raise questions whether the decisions influenced by opaque algorithms satisfy the requirement of reasoned awards, a cornerstone of arbitral legitimacy.
Transparency, Explainability, and Due Process
The main challenge of AI in the ISDS context is its explainability, since international investment laws always aims for procedural fairness and reason.[13] However, many AI systems particularly those based on machine learning operate as “black boxes,” making it difficult to trace how specific outcomes are reached.[14]
Tribunals may struggle to assess whether a given decision was arbitrary or discriminatory if the dispute comes from an AI driven regulatory mechanism as it will be very difficult to discuss the explanation of the algorithm’s function. Furthermore, the parties may ask for the disclosure of how these AI tools were used. And, this will raise the ultimate questions about evidence, confidentiality, and the burden of proof.[15]
There will therefore be huge problems ensuring the due process in the age of AI and may require new types of procedural safeguards. These could put pressure on states to maintain audit trails for algorithmic decisions, or guidelines for the permissible use of AI by tribunals and counsel. And, the ISDS legitimacy deficit (which is often associate with ISDS) can again be questioned without such safeguards.
Power Asymmetries and Global Inequality
There is a common fear that the integration of AI into ISDS also risks deepening existing power asymmetries between developed and developing states.[16] Technologically developed economies and multinational corporations are more likely to possess the technical expertise, data infrastructure, and financial resources needed to deploy sophisticated AI systems. Developing states, by contrast, may struggle to defend AI-related claims or to explain complex regulatory technologies before international tribunals.[17]
And, this will play a significant role for countries that already view ISDS as constrain to regulatory autonomy.[18] If AI driven governance becomes a source of liability under investment treaties, states may face increased exposure without corresponding capacity to manage legal risk. Equally, fear of ISDS claims may discourage states from adopting innovative AI based regulatory tools, undermining digital transformation and development goals.
Reforming ISDS for the Age of AI
AI can create new challenges for investor state dispute settlement (ISDS) mechanisms, so there is a need for ISDS reform.[19] There should be introduction of new treaty language where it will clearly define how investment protections can be applied if there is a use of right to regulation for public interest in context with the digital technologies. New generation investment treaties could incorporate provisions on transparency, data governance, and technological neutrality.
Furthermore, arbitral institutions at the procedural level may introduce guidelines on the use of AI in ISDS and how it will address the issues related with disclosure, accountability and human oversight.[20] There should be capacity building initiatives for the developing countries to ensure balance between developed and developing states during use of the AI in dispute settlement mechanisms.[21]
Finally, the aim should not be to resist technological change but to align AI with the fundamental values of international investment law: fairness, predictability, and legitimacy.
Conclusion
ISDS in the age of AI stands at a crossroads. Artificial intelligence has the potential to improve efficiency, consistency, and access to justice in investment arbitration, while also transforming the regulatory environments that give rise to disputes.[22] Yet, without careful governance, AI may exacerbate concerns about opacity, inequality, and democratic accountability that already plague the ISDS system.[23]
As AI becomes increasingly embedded in both state action and legal practice, ISDS must adapt. This adaptation requires not only technical understanding but also normative reflection on how algorithmic decision-making fits within the principles of international law.[24] The future legitimacy of ISDS may well depend on how successfully it navigates the promises and perils of artificial intelligence.
[1] Rudolf Dolzer and Christoph Schreuer, Principles of International Investment Law (2nd edn, OUP 2012).
[2] Gus Van Harten, Investment Treaty Arbitration and Public Law (OUP 2007).
[3] Ibid.
[4] UNCTAD, World Investment Report 2023: Investment and the Digital Economy (UNCTAD 2023).
[5] European Parliament & Council. (2024). Regulation (EU) 2024/… laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). This is the clearest evidence that AI is now formally embedded in a binding state regulatory framework. Use for strong doctrinal support. OECD. (2019). OECD Principles on Artificial Intelligence. Adopted by over 40 states and incorporated into national AI strategies and regulations. Good for international governance framing.
[6] OECD, Artificial Intelligence in Society (OECD Publishing 2019).
[7] European Commission, Ethics Guidelines for Trustworthy AI (2019).
[8] UNCTAD, World Investment Report 2023: Investment and the Digital Economy (UNCTAD 2023).
[9] Julian Arato, ‘The Logic of Contract in the World of Investment Treaties’ (2014) 58 William & Mary Law Review 351.
[10] Gabrielle Kaufmann-Kohler and Michele Potestà, ‘Can AI Replace Human Arbitrators?’ in Maxi Scherer (ed), International Arbitration and the Rule of Law (Kluwer Law International 2018).
[11] Anthea Roberts, Henrique Choer Moraes and Victor Ferguson, ‘Toward a Geographically Diverse, Machine-Learning Dataset of Investment Treaty Awards’ (2019) 19 Journal of International Economic Law 337.
[12] Gabrielle Kaufmann-Kohler and Michele Potestà, ‘Can AI Replace Human Arbitrators?’ in Maxi Scherer (ed), International Arbitration and the Rule of Law (Kluwer Law International 2018).
[13] Stephan W Schill, ‘Fair and Equitable Treatment under Investment Treaties as an Embodiment of the Rule of Law’ (2006) 51 ICSID Review 1.
[14] Cary Coglianese and David Lehr, ‘Regulating by Robot: Administrative Decision Making in the Machine-Learning Era’ (2017) 105 Georgetown Law Journal 1147.
[15] European Commission, Ethics Guidelines for Trustworthy AI (2019).
[16] UNCTAD, World Investment Report 2023: Investment and the Digital Economy (UNCTAD 2023).
[17] UNCITRAL, Report of Working Group III (Investor–State Dispute Settlement Reform), UN Doc A/CN.9/1004 (2019).
[18] Julian Arato, ‘The Logic of Contract in the World of Investment Treaties’ (2014) 58 William & Mary Law Review 351.
[19] UNCITRAL, Report of Working Group III (Investor–State Dispute Settlement Reform), UN Doc A/CN.9/1004 (2019).
[20] European Commission, Ethics Guidelines for Trustworthy AI (2019).
[21] UN General Assembly, Roadmap for Digital Cooperation (2020).
[22] UNCTAD, World Investment Report 2023: Investment and the Digital Economy (UNCTAD 2023).
[23] Gus Van Harten, Investment Treaty Arbitration and Public Law (OUP 2007).
[24] Rudolf Dolzer and Christoph Schreuer, Principles of International Investment Law (2nd edn, OUP 2012).
