AI on BBNJ High Seas Biodiversity Agreement for Legal teams at Ports & Terminals firms in international jurisdictions

Executive Summary

The BBNJ High Seas Biodiversity Agreement — formally the Agreement under the United Nations Convention on the Law of the Sea on the Conservation and Sustainable Use of Marine Biological Diversity of Areas Beyond National Jurisdiction — establishes a new multilateral governance layer over the high seas, including the authority of a Conference of the Parties to designate area-based management tools that may directly affect international shipping lanes.

Legal teams at Ports & Terminals firms in international jurisdictions are exposed to this treaty whenever their operations involve vessels transiting high-seas corridors that could fall within a designated area, and when advising on flag-state obligations, port-state controls, or bilateral shipping agreements that must be assessed against the treaty's non-undermining principle. Across the findings reviewed for this regulation, AI tools produced an error on a core provision governing COP authority over area-based management tools, misattributing the binding article reference in a way that would cause a legal team to locate and apply the wrong clause.

The failure pattern points to AI conflating general treaty-relationship provisions with the specific operative article, producing work-product that appears authoritative but is anchored to the wrong part of the text.

How AI gets this regulation wrong

AI tools tested on this regulation produced a specific kind of error: citing the correct legal principle but attributing it to the wrong article of the treaty, effectively pointing the reader toward a general clause rather than the provision that actually governs the authority in question. The table below breaks down the failure mode in detail, showing where AI invented a rule by confabulating article references.

What that means for your team

For Legal teams at Ports & Terminals firms, errors in treaty article attribution translate directly into flawed work-product: internal advice that cites the wrong provision, regulatory mapping documents built on a misidentified clause, or due-diligence memos that appear to reference the treaty accurately but would not withstand scrutiny from counsel familiar with the text. The table below maps each finding to the practical risk category it creates for the firm.

When this affects your department

Legal teams at Ports & Terminals firms are most likely to consult AI tools on the BBNJ Agreement when a vessel in their fleet operates on high-seas corridors, when management asks whether a newly designated marine protected area affects an existing service route, or when the firm must assess its obligations under flag-state or port-state frameworks that interact with the treaty.

The agreement's area-based management tool provisions are directly relevant to any route-risk or operational-compliance assessment involving the high seas, and Legal teams are often the first point of internal escalation when a maritime operations or commercial team asks whether the firm needs to alter a vessel's passage or port-call schedule.

The stakes are high because the BBNJ Agreement sits at the intersection of UNCLOS, IMO jurisdiction, and a novel COP-based governance structure. If Legal provides internal advice based on a wrong article reference — for instance, directing the firm to analyse its obligations under a general UNCLOS-relationship clause rather than the specific COP authority provision — the firm may reach a structurally correct conclusion through the wrong analytical path, leaving it unable to defend that advice if the underlying article is later scrutinised by external counsel, a flag-state regulator, or a port-state inspection authority.

In treaty interpretation, article-level precision matters: the scope of the non-undermining obligation differs depending on which provision it derives from, and a misattribution can determine whether the firm characterises a COP area-based management decision as subject to challenge or as binding.

Beyond route and vessel compliance, the BBNJ Agreement has downstream implications for terminal operators with capacity in ports adjacent to high-seas corridors, for firms involved in transhipment hubs, and for Legal teams advising on charter-party negotiations where force majeure or regulatory-change clauses must account for emerging treaty obligations. In each of these contexts, the Legal team's output — whether a written memo, a contractual provision, or a regulatory mapping schedule — is the firm's formal position, and an AI-introduced article error that goes unchecked becomes part of the record.

The findings at a glance

The table below summarises each finding identified for this regulation, covering the question area tested, the nature of the AI error, and the risk category it creates for a Legal team at a Ports & Terminals firm in international jurisdictions.

Aggregate impact

The finding reviewed for this regulation reveals a pattern of article-level confabulation: AI tools correctly identify the substantive principle — in this case, the non-undermining obligation that constrains COP authority over area-based management tools in international shipping lanes — but anchor that principle to the wrong article of the treaty. Article 22(2) of the BBNJ Agreement is the provision that specifically governs COP decisions on area-based management tools and their relationship to existing legal instruments and bodies such as the IMO. AI tools attributed this constraint to Article 8, which addresses the general relationship between the Agreement and UNCLOS.

The distinction is not trivial: Article 22(2) sets the standard applicable to a specific class of COP decisions, while Article 8 operates at the level of the agreement as a whole.

For a Legal team, this error clusters on the exact question that matters most to a Ports & Terminals firm: whether and how the COP can act in a way that affects vessel transit rights in a shipping lane. A Legal team that uses an AI-generated article reference without verifying it against the treaty text will produce work-product that points to the correct conclusion — the COP must respect IMO competence — but cites the wrong legal authority.

That structural error exposes the firm to a specific risk: if the advice is ever challenged, external counsel or a regulatory body reviewing the firm's legal position will identify the misattribution and the credibility of the entire analysis will be undermined.

The systemic risk here is compounded by the novelty of the BBNJ Agreement. Because the treaty entered into force recently and its implementing mechanisms are still developing, there is limited secondary commentary and jurisprudence to cross-check against. Legal teams that rely on AI tools to navigate new treaty text — precisely the situation where primary-source verification is most important — face a heightened risk that article-level errors will go undetected until they surface in an adversarial context.

What your team should do

The default position for Legal teams at Ports & Terminals firms should be to treat any AI-generated reference to a specific BBNJ Agreement article as unverified until it has been checked against the treaty text directly. The United Nations Treaty Collection publishes the full treaty text at treaties.un.org, and article-level verification takes minutes.

Given the finding reviewed here — where AI correctly stated the operative principle but cited the wrong article — the practical safeguard is not to distrust the AI's characterisation of the law, but to verify that the article it names actually contains the words it attributes to it.

For Legal teams, AI tools are reasonably safe for orientation tasks on the BBNJ Agreement: summarising the treaty's general architecture, identifying the broad categories of COP authority, flagging which international bodies (IMO, ISA, regional fisheries organisations) are referenced, or generating a first-draft list of questions for external maritime law counsel. Where AI tools should not be trusted without verification is in any task that requires article-level citation precision — internal legal memos, compliance schedules, due-diligence reports, or contractual provisions that reference specific treaty obligations. In those contexts, the article number the AI provides must be checked, not assumed.

A practical workflow for this regulation is to use AI tools to identify the relevant part of the treaty (for example, Part III on area-based management tools) and then read that part of the treaty directly before committing an article reference to any written output. For questions involving the interaction between BBNJ COP authority and IMO jurisdiction — the most operationally sensitive question for a Ports & Terminals firm — the relevant provision is Article 22(2), not the general UNCLOS-relationship clause.

Legal teams that build this one verification habit into their AI-assisted workflow on this treaty will avoid the specific error pattern identified in this cell.

How RLB Can Help

RegLeg's published Hallucination Research gives the Legal team at a Ports & Terminals firm a free, structured pre-flight check before placing weight on AI-assisted regulatory analysis. Where an AI tool has been tested against a relevant regulation — port state control regimes, customs and trade facilitation rules, environmental discharge standards, or cross-border cargo liability frameworks — the research shows specifically where the tool produced confident but incorrect outputs. Legal counsel can consult these findings before citing AI-generated regulatory summaries in contracts, compliance opinions, or regulatory submissions, reducing the risk of acting on a hallucinated rule or an outdated statutory reference.

Beyond the published research, RLB works with Legal teams to map which AI-supported workflows in a Ports & Terminals function carry the highest hallucination exposure. Port and terminal operations touch a layered regulatory environment — IMO conventions, flag-state and port-state requirements, bilateral trade agreements, and jurisdiction-specific port authority directives — where AI tools frequently conflate similar-sounding instruments or apply requirements from one jurisdiction to another.

A bespoke regulator deep-dive identifies the specific regulatory domains and workflow steps where that exposure is greatest, giving Legal a prioritised view of where human verification is non-negotiable and where AI assistance can be relied upon with appropriate checks.

RLB also offers a confidential review of the firm's existing AI-use policy against our failure-mode catalogue, with prioritised remediation recommendations tailored to a Legal function's risk profile. For teams building internal capability, we can develop training material and CPD-aligned content that equips lawyers and compliance professionals to interrogate AI output critically — understanding the classes of error AI tools are prone to in regulatory contexts, and applying consistent verification disciplines without needing to audit every output from first principles.