Artificial Intelligence Under Control
Artificial Intelligence Under Control
A Critical Look at the Regulation of the Law Promoting Its Use
In Peru, Law No. 31814, published on July 5, 2023, and its Regulation, approved by Supreme Decree No. 115-2025-PCM, published on September 9, 2025, constitute the legal framework aimed at public entities, private organizations, civil societies, and others, with the purpose of promoting the safe, responsible, and ethical use of Artificial Intelligence (AI)-based systems, as well as establishing obligations for those who develop or implement AI systems, setting out principles and controls that must be followed according to their risks and uses.
What is meant by an implementing company and a developing company?
- An implementing company is one that incorporates AI into its processes.
- For example, a supplier tool that extracts data from invoices or automates processes.
- A developing company is one that designs, programs, or trains its own system.
- For example, the company designs and programs an AI system and trains it with product images to identify manufacturing defects in order to offer the system to industrial companies.
Both must comply with obligations and assume responsibilities based on their own risks.
Types of Risks from the Use of AI-Based Systems
- (a) Misuse: Any AI-based system used in a way that irreversibly, significantly, and negatively impacts fundamental rights or people’s well-being under any circumstances. Such use is therefore prohibited.
- (b) High-risk use: Any AI-based system whose use poses a risk to human life, dignity, freedom, physical safety, and other fundamental rights, considering both the intended objective and possible unintended consequences.
- (c) Acceptable risk: Any risk not included in the previous classifications.
When does the Regulation become applicable and what are the consequences of non-compliance?
The Regulation is already in force, but management and transparency obligations will be implemented progressively according to the use of AI in different sectors:
| Sectors | Deadline |
|---|---|
| Health, education, justice, security, economy and finance | September 2026 |
| Transport, trade and labor | September 2027 |
| Production, agriculture, energy and mining | September 2028 |
| Other uses | September 2029 |
Although the Law and its Regulation do not establish their own fines, powers are granted to the Presidency of the Council of Ministers to supervise and communicate possible violations to the competent authorities. Likewise, any person may report misuse through gob.pe/iaperu or file a complaint with INDECOPI or the National Authority for Personal Data Protection.
Important: The Regulation excludes from its scope the use of AI-based systems for personal purposes and those used for national defense and security, provided that certain fundamental principles are met.
Did you know…?
On October 21, 2026, the Secretariat for Government and Digital Transformation approved Directive No. 002-2026-PCM/SGTD, which establishes the profile and responsibilities of the Artificial Intelligence Officer, with the purpose of enabling public entities to consider criteria for the profile, appointment, and definition of responsibilities of the Artificial Intelligence Officer. The Directive addresses obligations according to the different levels of government and their existing maturity in data management and digital systems.
Private entities may use the criteria established in this Directive as a reference whenever it generates value for them.
But… What is currently missing?
- No specific sanctions are established for cases of non-compliance related to personal data breaches, consumer rights violations, or other infringements that may arise from the use of AI systems.
- The Regulation refers to tools such as impact assessments and algorithmic transparency mechanisms, but it does not provide detailed guidelines or concrete implementation criteria.
- Greater obligations are imposed on the public sector than on the private sector. Strictly speaking, carrying out impact assessments appears to be largely left to the discretion of private companies, even though such assessments are a fundamental element in determining the level of risk associated with the use of AI and, consequently, the control measures that should be applied.
This regulatory difference could create a gap in the effective management of the risks represented by these technologies, because some biases or adverse effects may not be identified in a timely manner, and corrective measures may be adopted only after the system is already affecting people.
Our Recommendations
- Identify the AI systems used in operations and classify them according to the level of risk they represent.
- Approve internal policies and rules for the responsible use of AI, complemented by training programs for personnel.
- Review contracts signed with providers of these technologies in order to define responsibilities and risk-allocation mechanisms for possible errors, failures, or damages arising from the operation of AI systems.
- Evaluate the need to appoint an Artificial Intelligence Officer based on the contribution the company expects AI to make toward achieving its strategic objectives. On that basis, analyze current and planned AI uses, their risks, and the organization’s maturity level in data management and digital platforms to determine whether such a role is necessary to guide responsible adoption aligned with the organization’s expected outcomes
Contact Us
- Carina Dávila, Legal Director - Email: cdavila@bdo.com.pe
- Maria Luisa Pérez, Legal Advisor - Email: mperezb@bdo.com.pe