Loading...
- Home
- Capabilities
- ACTIO® Hub
- About Us
- Connect with Us
- AP Library
June 3, 2026
Originally published in ACTIO E-Magazine Edition 29 (June 2026). Law as of June 2026.
Yes, arbitrators may use AI as an assistant, but not as a substitute decision-maker. Under Law No. 30 of 1999 on Arbitration and Alternative Dispute Resolution, the adjudicative function is a personal mandate entrusted to a human decision-maker, awards must contain reasons, parties must be treated equally, and proceedings are confidential. AI may help with research, review, and drafting, but arbitrators must independently evaluate evidence, verify authorities, and exercise personal judgment.
Artificial intelligence has rapidly evolved from experimental technology into a widely used professional tool, including in legal and adjudicative contexts. In international arbitration, AI systems are now used to summarise submissions, identify precedents, review voluminous evidence, and even generate draft procedural orders or sections of awards. While these applications promise efficiency and cost savings, they also raise profound questions about adjudicative integrity.
Globally, arbitral institutions such as the ICC and SIAC have begun examining the ethical boundaries of AI use. The Silicon Valley Arbitration & Mediation Center (SVAMC) issued Guidelines on the Use of Artificial Intelligence in Arbitration (2024), which emphasise human responsibility and transparency in AI-assisted processes. Meanwhile, courts in multiple jurisdictions have addressed AI-related misconduct by counsel, including the citation of non-existent or fabricated authorities in pleadings.
The debate intensified following LaPaglia v. Valve Corp., where a party sought to vacate an arbitral award on the basis that it allegedly relied on non-existent legal authorities, purportedly generated by AI. Citations of facts that were untrue and were not presented before the hearing were found. Although the U.S. court refused to vacate the award, the case prompted broader discussion about whether the use of AI by arbitrators constitutes impermissible delegation of decision-making.
In Indonesia, Law No. 30/1999 does not expressly regulate AI, but it embeds foundational principles: party autonomy, equal treatment, confidentiality, and the requirement that awards contain reasons. Annulment is narrowly confined to fraud, forged documents, or concealed decisive documents (Article 70), reflecting a philosophy of minimal court intervention and strong finality. Against this backdrop, the central question is not whether AI may be used, but where assistance ends and unlawful delegation begins.
Arbitrators are appointed based on expertise and independence. The adjudicative function is inherently entrusted to a specific human decision-maker, even though tribunals routinely rely on secretaries for administrative support. If AI is used for grammar checks, document organisation, or preliminary research, it resembles traditional clerical assistance. However, if an arbitrator were to rely blindly on AI-generated reasoning or legal authorities without independent verification, the concern shifts from assistance to delegation.
Under Law No. 30/1999, while there is no explicit prohibition against technological assistance, the personal nature of the mandate is implicit in Articles 12-15, which regulate arbitrator qualifications and independence. Delegating reasoning to an opaque algorithm may undermine this personal trust relationship.
Article 29(1) mandates that parties are treated equally and given a full opportunity to present their case. If a tribunal relies on AI-generated authorities not introduced into the record, including potentially fabricated citations, the parties may be deprived of their right to comment, raising serious due process concerns. Comparatively, Article V(1)(b) of the New York Convention permits refusal of enforcement where a party was unable to present its case. Indonesian courts interpret this exception narrowly, but a demonstrable denial of due process could engage the public policy (ketertiban umum) defence under Article 66 of Law No. 30/1999.
Article 54(1) requires arbitral awards to contain “the arbitrator’s or arbitral tribunal’s consideration and conclusion regarding the entire dispute”, and Article 56(1) provides that the arbitrator or tribunal shall make a decision based on the provisions of law or based on justice and propriety. AI systems are known to generate plausible but sometimes inaccurate outputs (“hallucinations”), which may affect the reliability of reasoning where such outputs are not independently verified. An award based on non-existent authorities and fictitious citations may not readily qualify as “provisions of the law” within Article 56(1), particularly where the underlying sources cannot be substantiated. However, unless such reliance falls within fraud or concealed documents under Article 70, annulment may remain unavailable, suggesting a potential gap between existing annulment grounds and emerging risks. A more immediate risk may arise at the enforcement stage, particularly in cross-border contexts where courts may scrutinise the integrity of the decision-making process more closely.
Arbitration in Indonesia is confidential (Article 27). Uploading submissions to public AI platforms risks disclosure of sensitive data, which may be particularly consequential given the confidential nature of arbitral proceedings. Furthermore, Law No. 27 of 2022 concerning Personal Data Protection, as amended by Law No. 1 of 2026 concerning Criminal Adjustment, imposes obligations on data controllers, including cross-border data transfer safeguards, which may require careful consideration when using third-party AI systems.
If arbitrators input confidential materials into unsecured AI systems hosted abroad, questions arise as to compliance with data protection and confidentiality obligations. Such conduct may expose institutions and arbitrators to reputational and potential legal risks.
In LaPaglia, the Court dismissed the petition to vacate the arbitral award on jurisdictional grounds, without addressing the substantive AI-related allegations. The case nevertheless highlights emerging questions regarding the use of AI in arbitral decision-making. From an Indonesian perspective, it illustrates two key considerations: first, annulment remains exceptional, and Article 70 would not easily capture AI-related reasoning errors unless linked to fraud or concealed decisive documents; second, the legitimacy of arbitration depends less on technological prohibition and more on maintaining human accountability. The legal limits under Law No. 30/1999 may therefore be understood as follows: AI may assist, but arbitrators must independently evaluate evidence, verify authorities, and exercise personal judgment.
May an arbitrator use AI at all under Indonesian law?
Yes. Law No. 30/1999 contains no explicit prohibition against technological assistance. The limit is that AI must assist, not decide: the arbitrator must independently evaluate evidence, verify authorities, and exercise personal judgment.
What is the difference between assistance and delegation?
Assistance includes grammar checks, document organisation, and preliminary research. Delegation arises where an arbitrator relies blindly on AI-generated reasoning or legal authorities without independent verification.
Could an AI-assisted award be annulled in Indonesia?
Not easily. Annulment under Article 70 is confined to fraud, forged documents, or concealed decisive documents, so AI reasoning errors alone would not readily fall within those grounds unless linked to fraud or concealed documents.
What is the enforcement risk?
At the enforcement stage, particularly cross-border, courts may scrutinise the integrity of the decision-making process. Under Article V(1)(b) of the New York Convention, enforcement may be refused where a party was unable to present its case, and a denial of due process could engage the public policy defence under Article 66.
Why is confidentiality a particular concern?
Because arbitration is confidential under Article 27, and uploading submissions to public AI platforms risks disclosure. The PDP Law also imposes obligations on data controllers, including cross-border data transfer safeguards.
The autonomous era presents legal systems with the challenge of balancing efficiency and fundamental legal values. Arbitration, founded on trust and deliberative judgment, requires that its human decision-making function remain central. Indonesia’s readiness will not be measured by how quickly AI is adopted, but by how carefully its limits are defined. Delegation undermines legitimacy; responsible assistance enhances it.
