Governance

Accountability (OECD)

AI actors are responsible for proper system functioning.

How the laws define this

Core consensus establishes accountability as a requirement to assign clear responsibility, maintain documentation, and implement oversight or corrective measures so actors remain answerable for AI systems and their outcomes. Every approach centers on making system actions attributable to specific roles or organizations. However, the instruments diverge significantly regarding enforcement mechanisms and organizational focus. The Artificial Intelligence Regulation Bill (Legislative Assembly) takes a punitive approach by explicitly tying accountability to administrative or criminal liability for misconduct. Conversely, the Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of AI and Data Analytics (MAS) and the Artificial Intelligence (AI) Ethics Guidelines for Digital Thailand focus on internal governance infrastructure, prescribing specific roles like board oversight, ethics committees, and escalation or redress channels. Distantly, the Artificial Intelligence (AI) Technologies and Recordkeeping Policy frames accountability around technical recordkeeping, requiring processes, inputs, and outputs to be sufficiently documented and understandable to allow for external scrutiny.

Synthesised from the 4 statutory definitions below. Each one is quoted in full, with its source.

How jurisdictions differ

What is distinctive in each definition, binding law in force first.

Jurisdiction	Instrument	Status	What is distinctive
Singapore	Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of AI and Data Analytics (MAS)	In Force	Adds explicit governance structures like model owners, independent validators, board oversight, clear escalation channels, and oversight across the AIDA lifecycle.
Victoria, Australia	Artificial Intelligence (AI) Technologies and Recordkeeping Policy	In Force	Limits accountability to sufficient documentation and understandability of AI processes, inputs, and outputs for scrutiny and responsibility attribution.
Thailand	Artificial Intelligence (AI) Ethics Guidelines for Digital Thailand	Adopted	Adds procedures for audit, redress, and escalation alongside governance structures like ethics committees to hold actors responsible for harms.
Costa Rica	Law Regulating Artificial Intelligence in Costa Rica	Under Review	Adds obligations for documentation, corrective measures, and exposure to administrative or criminal liability for organizations deploying AI systems.

What it triggers

In the Law Regulating Artificial Intelligence in Costa Rica, organizations deploying AI systems would be subject to administrative or criminal liability for misconduct.source

Definitions (4)

The obligation for organizations deploying AI systems to establish clear responsibility chains and take responsibility for AI outcomes, including maintaining documentation, implementing corrective measures, and being subject to administrative or criminal liability for misconduct.

Artificial Intelligence Regulation Bill (Legislative Assembly)Definition 1 of 4

Accountability requires explicit governance structures and role‑based responsibilities (such as model owners, independent validators and board oversight), clear escalation channels, and mechanisms to ensure oversight and remedial action across the AIDA lifecycle.

Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of AI and Data Analytics (MAS)Definition 2 of 4

The assignment of clear roles, accountable officers and governance structures (e.g., ethics committees) together with procedures for audit, redress, and escalation so that actors can be held responsible for design, deployment and harms arising from AI systems.

Artificial Intelligence (AI) Ethics Guidelines for Digital ThailandDefinition 3 of 4

A central tenet, implying that the processes, inputs, and outputs of AI systems must be sufficiently documented and understandable to allow for scrutiny and attribution of responsibility.

Artificial Intelligence (AI) Technologies and Recordkeeping PolicyDefinition 4 of 4

Not to be confused with

Accountability Mechanisms

Accountability mechanisms focus on responsibility chains and redress for harms, while accountability here encompasses broader governance, documentation, and liability aspects.

shared responsibility

Shared responsibility emphasizes distributed duties across stakeholders, whereas accountability centers on clear roles, structures, and individual organizational obligations.

Questions and answers

What governance is needed for AI accountability under Singapore FEAT Principles?

Explicit governance structures and role-based responsibilities such as model owners, independent validators and board oversight, clear escalation channels, and mechanisms for oversight and remedial action across the AIDA lifecycle. These are non-binding guidelines from MAS.

What liability applies to AI under Costa Rica's bill?

Organizations deploying AI systems would need to establish clear responsibility chains, maintain documentation, implement corrective measures, and face administrative or criminal liability for misconduct. The bill is under review.