Chile - AI Development Limits (17112-19)

Overview

Boletín 17112-19, a parliamentary motion introduced in the Chamber of Deputies on 3 September 2024, seeks to establish statutory limits on AI design, development and operation to protect fundamental human rights. The motion’s stated object is to "ensure full respect for the fundamental rights of human beings by establishing the indispensable limitations to prevent harmful or undesired effects" caused by AI systems. The initiative frames AI regulation as a rights protection exercise and introduces a risk-oriented compliance model that places duties on developers, implementers and deployers of AI systems. The bill was presented to the Cámara’s relevant commissions for hearings and is recorded in official parliamentary tramitación pages and session agendas; see the Chamber's project page and session citations for source material. For the official tramitación record consult Chamber of Deputies - Tramitación (Boletín 17112-19).

Definitions

The bill defines core terms to delimit scope and obligations: "intelligent system" / "AI system" (automated or semi-automated decision systems using data-driven models or algorithmic logic to infer, predict, classify or generate outputs affecting human decisions or social outcomes); "developer" (natural or legal persons responsible for model creation and training); "operator/implementer" (entities deploying, integrating or providing use of AI systems in Chile); "controller" (entities determining the purposes and means of processing); and "incident" (adverse effect or material risk to fundamental rights arising from AI operation). These definitions are designed to apply extraterritorially where outputs of systems are used or have effects in Chile. The proposal also sets out a risk taxonomy (unacceptable, high, limited, minimal) to calibrate duties and limitations.

Governance and Institutional Framework

The motion assigns supervisory and advisory responsibilities to existing and prospective public bodies. The text contemplates coordination with the Ministry of Science, Technology, Knowledge and Innovation as a lead policy actor for technical oversight and coordination; the bill also contemplates the creation of a consultative technical advisory council on AI to propose lists of high-risk systems, issue technical guidance, and review law implementation. Parliamentary committee records show that the Ministry of Science has been invited to hearings, indicating anticipated executive-legislative coordination; see the Chamber citations for the relevant commission sessions and invited testimony at Cámara - Commission citation (18 June 2025). Enforcement arrangements contemplate administrative oversight, market surveillance and cooperation with data protection authorities when personal data processing is implicated. The proposed governance model aims to combine technical expertise (via advisory council) with regulatory authority located in the Executive branch, subject to parliamentary oversight.

Key Focus Areas

The bill concentrates on several interlocking priorities: (1) fundamental rights protection — explicitly naming autonomy, privacy, transparency, security and the right to remedies — and establishing them as guiding principles; (2) risk classification — a tiered approach (unacceptable, high, limited, minimal risk) to determine the stringency of obligations; (3) mandatory risk and impact assessments before deployment of high-risk systems; (4) documentation, logging and model cards/technical files to support auditability and traceability; (5) human oversight for systems affecting rights; (6) data protection and privacy-preserving measures when training and operating AI systems; (7) incident reporting and mandatory remediation when adverse impacts occur; and (8) accountability and liability rules to ensure reparations for victims. The bill also contemplates prohibiting or restricting certain uses of AI deemed inherently incompatible with rights protection.

Implementation Framework

Implementation relies on a mix of mandatory technical documentation requirements, pre-deployment risk assessments, registration or notification duties for specified categories of systems, and post-market monitoring. Developers and implementers would be required to prepare and retain compliance packs (technical documentation, risk assessment, data provenance logs, evidence of human oversight mechanisms and security measures). For high-risk systems the proposal contemplates more stringent obligations (deeper testing, third-party auditability and explicit prior notification to the designated authority). The bill foresees transitional and support measures to avoid undue burden on small actors and to foster innovation while ensuring rights protection; legislative discussions and commission agendas reference the need for calibrated support and linkage with other public policy initiatives such as national AI action plans; see Ministry of Science — AI regulatory overview.

Monitoring and Evaluation

Monitoring combines ex ante and ex post instruments: mandatory registration/notification for selected systems, periodic reassessments, incident reporting obligations, and authority-led market surveillance including audits and spot checks. The bill envisages periodic public reporting and an evaluation cycle to review lists of high-risk systems and to update technical guidance every three years (as a model of dynamic regulatory governance). Parliamentary commission hearings indicate an intent to integrate stakeholder inputs and technical expertise into monitoring processes, including invitations to industry representatives and legal experts for evidence sessions.

Penalties, Liability, and Appeals

The motion establishes that developers, operators and controllers are responsible for damages caused by AI systems and must repair harms, including unanticipated harms causally linked to system decisions. Administrative penalties (fines, orders to suspend or withdraw systems from service) are foreseen for non-compliance; in cases of severe negligence or willful misconduct, civil or criminal liability may apply under existing Chilean law. The bill also contemplates administrative appeal routes and judicial remedies for affected individuals seeking redress. Public summaries of the motion emphasize reparatory duties and chain-of-responsibility principles for legal accountability.

Relationship to Other Instruments

Boletín 17112-19 sits alongside other national initiatives on AI and related rights: the Executive’s regulatory project on AI (Boletín 16821-19 / 16.821–19) and parliamentary proposals on neuro-rights (Boletín 13828-19-S). Legislative records show discussion about reconciling overlapping aims and potentially consolidating or coordinating provisions to avoid fragmentation; committee transcripts include debate on whether motioned bills should be subsumed or harmonised. The bill also interfaces with Chilean data protection frameworks and existing sectoral regulation (health, finance, public procurement), requiring coordination with sector regulators and data protection authorities (where personal data is processed).

International Alignment

While domestically focused, the motion reflects international regulatory trends: rights-based, risk-tiered frameworks similar to the EU's AI Act proposals and other national approaches emphasizing high-risk system obligations, transparency and accountability. The drafters and parliamentary debates reference international standards and cooperation modalities; Chilean authorities have signalled interest in cooperation with foreign partners on AI governance and infrastructure. The bill’s extraterritorial reach (applying to systems whose outputs are used in Chile) is consistent with global regulatory practices for digital services and ensures imports and remote providers can be regulated in relation to effects on Chilean residents.

Implementation Timeline

Sources and References