Colin Ong(Brunei)

President of the Arbitration Association Brunei Darussalam (AABD)

Chairman of the International Advisory Board of the Thailand Arbitration Center (THAC)

General introduction to modern commercial arbitration 

A. Parties who are involved in long term or complex high value contracts need to be able to resolve any legal disputes in a way that preserves the parties ability to continue performing the contract without failure or undue delay and costs despite any dispute.

B. There are major differences in legal regimes practised across the BRI region. These include common law (examples being the South Asian countries and also South East Asian countries like Malaysia and Singapore), civil law (found in Central Asia) and Islamic law (in the Middle East). Both Chinese and non-Chinese contractors may commit to projects that carry significant delay risk. Governments may change half way through construction. Because balances of power shift, governments change and new policies come into play. 

C. International Commercial Arbitration enables contracting parties to settle disputes confidentially, through neutral national arbitrators in order to reach final and binding decisions. 

D. Nothing is as important in international commercial arbitration than to safeguard trust, fairness and to ensure a good business environment between the parties so that long term projects can be performed smoothly and successfully into the future. 

E. All arbitral awards must be enforceable in different countries that are signatories to the New York Convention. The New York Convention has been described by the UNCITRAL Secretary as “one of the most important and successful United Nations treaties in the area of international trade law, and the cornerstone of the international arbitration system”. The New York Convention has 175 contracting states as of 2026.

Impact of generative artificial intelligence on modern commercial arbitration 

There are very few modern developments in commercial arbitration that may have a significant impact on the applicability of the New York Convention. The most important development is the advancement, use and abuse of Generative AI in commercial arbitration.

Generative AI now offers increased efficiency in traditional work scope of filtering documents; document analysis, suggested AI drafting, and case management. Before the advent of Generative AI, lawyers were only able to summarise key points of complex and voluminous documents by actually physically reading through the documents themselves. 

Generative AI can draft initial drafts of pleadings; analyse opposing submissions, and even identify fact patterns in past cases and the likelihood of probable decisions by tribunals. Generative AI algorithms can quickly summarise lengthy documents and identify the key points of the same. Generative AI allows for almost instantaneous comparison of opposing arguments across different documents. This creates efficiency  which generally translates into much costs savings by parties. 

While Generative AI offers clear benefits in efficiency and service delivery, it also raises serious ethical concerns. These include misinformation and even the phenomenon of hallucination where AI invents fictitious cases and laws. “Hallucinations” often occur in the industry, because the models generate responses based on statistical patterns learned from large datasets rather than by verifying the actual facts in those datasets. AI tools raise concerns about transparency, bias, accuracy, confidentiality and due process.

Generative AI is capable of producing content that is highly convincing, yet completely fabricated. Hallucinated legal texts often appear stylistically legitimate, formatted with citations, statutes and judicial opinions, creating a false illusion of credibility that can mislead even experienced legal professionals. This makes it difficult to distinguish real information from fake and it poses a serious risk to parties and arbitral tribunals alike. Deepfake documents are fabricated audiovisual content that have been created or altered to appear to be genuine and authentic in the eyes of viewers. The danger of deepfake documents has the potential to completely distort the truth of evidence presented. In the worst situations, deepfakes and hallucinations will lead to injustice, distort and alter the outcome of arbitrations.

The High Court of England and Wales has made it clear that the duty to verify the accuracy of AI-based research falls on lawyers who either use AI to conduct research themselves, or who rely on such research done by others. It is akin to the duty of a lawyer to bear responsibility for work carried out by a trainee solicitor; a pupil barrister or a relatively junior barrister.

The Singapore High Court decision of Tajudin v Suriaya [2025] SGHCR 33 addresses the improper use of generative AI in legal submissions, marking one of Singapore’s early decisions to penalise counsel for AI hallucination specifically, the citation of fictitious legal authority. 

The Singapore High Court cited the English judgment of Ayinde v The London Borough of Haringey [2025] EWHC 1383 as part of a troubling global trend of relying on unverified AI-generated authorities in legal proceedings. The Singapore HC emphasises that lawyers remain personally responsible for verifying the accuracy of all submissions generated with AI tools.

Arbitrators and lawyers alike must grapple with questions of authorship, accountability, and transparency in Generative AI-generated outputs. The failure by a tribunal to follow due process is the most common ground to challenge an arbitral award before the national courts. In the event that the parties did not consent to the tribunal’s use of Generative AI in the arbitration proceedings, this will create an opportunity for the unsuccessful party to challenge an award based on procedural irregularity arising from the use of AI. Equally, a party that wishes to delay the proceedings may also bring a challenge against the tribunal for using generative AI, without disclosure and consent being sought by the tribunal.

The arbitral community has yet to work out standard universal guidelines on how to use Generative AI responsibly. Despite differences across the various guidelines, a clear consensus has emerged that AI may support arbitral proceedings but human responsibility, oversight, and verification remain paramount, and decision-making must not be delegated to AI. This broadly reflects the approach taken by national courts in England and Wales and Singapore.

The New York Convention is now being put to a test its drafters never anticipated in 1958.

Article V(1)(d) of the NYC may be triggered in a situation where a human arbitrator rubber-stamps an AI-generated award without independently evaluating the merits. This would amount to a dereliction of an arbitrators duty as the parties expected a human judgment they had contracted for. It would amount to a procedural irregularity under the Convention.

Enforcement of an award may also be resisted under Article V(1)(b) in the event that AI tools produce opaque outputs without adequate disclosure to the Parties.

In addition, Article V(2)(b) would also be triggered where AI introduces demonstrable systemic bias that materially affects the outcome.

Across major jurisdictions, arbitration legislation was written with human decision-makers in mind. The laws do not clearly authorise a robot to act as arbitrator, so human appointment remains the default. In Australia, the International Arbitration Act 1974 incorporates the UNCITRAL Model Law and uses the word “person” more than 40 times, including in section 18A on the appointment of arbitrators. The Cayman Islands Arbitration Law defines an arbitrator as a member of the tribunal and section 18(1) refers to when “a person is approached” for possible appointment, making it clear that a robot cannot serve as an arbitrator. Indonesia defines arbitrators as “one or more persons” appointed by the parties, the court or an institution.

Article 11 of the UNCITRAL Model Law provides that no person shall be precluded by nationality from acting as an arbitrator and the use of “his” and “no person” confirms that the drafters intended a human arbitrator.  Japan’s law also uses person-based terms, and Singapore’s Arbitration Act 2001 refers to “person” more than 40 times, including section 13(1), which confirms that no person is precluded by nationality from acting as an arbitrator.

As it was put together in 1958, the New York Convention cannot be easily redrafted to address AI-assisted awards without the agreement of the 175 signatory countries.

Mere ratification of the New York Convention does not waive state immunity

There have been recent judicial decisions including the English Court of Appeal in CC/Devas (Mauritius) Ltd & others v The Republic of India [2026] EWCA Civ 797 and the Australian High Court in CCDM Holdings LLC & Ors v The Republic of India [2026] HCA 9 which both confirmed that a state’s ratification of the New York Convention does not amount to a waiver of state immunity.

The Courts in both of the common law jurisdictions held that clear words would be needed to override the principle of sovereign immunity, and no such words are found in Article III of the New York Convention. It is expected that future civil law courts would also be likely to take the same approach if faced with the same challenge.

Environmental, social, and governance (“ESG”) and climate-related arbitrations

ESG issues continue to emerge as a key feature in international commercial arbitration, especially relating to infrastructure, energy and natural resources. 

Public policy defences to enforcement of awards under Article V(2)(b) of the NYC continue to be raised by States who justify enforcement of legislation to protect safety of their people and preservation of energy and natural resources

Sanctions are increasingly used to justify non-enforcement of contracts and awards

Sanctions against Russia have disrupted contracts and resulted in numerous investor-State disputes involving foreign and Russian investors. In the context of arbitration in the United States, permission must be obtained in arbitration involving sanctioned persons or entities, by way of a license from the Office of Foreign Asset Control (OFAC). 

Similarly, persons from certain European countries are not entitled to sit as arbitrators unless they obtain license from their own countries. Lawyers and arbitrators from quite a number of nationalities tend to decline getting involved in cases involving sanctioned Russian entities so as not to potentially fall in breach of sanctions imposed by their own countries. Russian courts refuse to enforce validly rendered arbitral awards from seats located in countries that have imposed sanctions on Russia.

Not only are Russian courts refusing to enforce validly rendered arbitral awards but courts of other countries have also refused to enforce arbitral awards rendered in Russia. In two separate decisions issued in 2025, higher courts in Germany (in Stuttgart and Frankfurt) declined to recognise and enforce Russian arbitral awards on the ground that enforcement would contravene EU sanctions against Russia. 

This trend is expected to continue from the reintroduction of sanctions against Iran. US President Executive Order 13902 of 10 January 2020 is an example of which secondary sanction may be imposed upon non-US individuals or entities if they operate in or knowingly engage ‘in a significant transaction for the sale, supply, or transfer to or from Iran of significant goods or services used in connection with a sector of the Iranian economy’ specified in the E.O

As a conclusion, all law firms and lawyers across the world will need to be diligent and must understand and respect sanctions imposed by any major country. If they do not understand or respect such sanctions laws, the lawyers themselves can potentially be sued for professional negligence if the award can be challenged and rendered unenforceable under Article V(2)(b) for being in breach of which ever sanction is applicable.