Many companies use automated models to decide who is granted credit, which candidate moves on to an interview or which customer is flagged as a fraud risk. Until 2025, Mexican personal data law said nothing specific about those decisions. The law in force does, and adapting to it requires redesigning processes as well as documents.
What changes for your company
The new ground for objection. Article 26, section II, of the Federal Law on the Protection of Personal Data Held by Private Parties (Ley Federal de Protección de Datos Personales en Posesión de los Particulares) allows the data subject to object to the processing of their data, or to demand that it cease, when the data is subject to automated processing that produces unwanted effects for them or significantly affects their interests, rights or freedoms, and that is intended to evaluate personal aspects without human intervention or to analyze or predict their professional performance, economic situation, health, sexual preferences, reliability or behavior.
Which decisions fall within it. The provision requires three elements to be present together: automated processing, a significant effect on the individual and the evaluation of personal aspects without human intervention. A model that automatically denies credit, rejects a job application or blocks an account based on a risk score meets, in principle, all three. By contrast, a tool that only organizes information so that a person can decide, with real ability to change the outcome, is unlikely to meet them.
What the law does not say. The 2025 law does not establish a general right to obtain an explanation of the algorithm or to demand human review as such, nor does it require impact assessments before a model is used. It grants the right to object, which is processed under the procedure and deadlines for the rights of access, rectification, cancellation and objection: a response within a maximum of twenty business days and implementation within the following fifteen, extendable once (Article 31). Even so, to handle an objection without halting the business, the soundest approach is to have an effective channel for human review.
The limit of the right. Objection is not available when the processing is necessary to fulfill a legal obligation imposed on the data controller (Article 26, last paragraph). An automated process that the company carries out by mandate of law, for example to identify customers in compliance with anti-money laundering rules (prevention of transactions with funds of illicit origin), may fall outside the right, but that necessity must be demonstrable.
What the notice must disclose. If the company uses data for automated evaluations, that purpose must be stated in the privacy notice, and the notice must indicate the means to exercise the right to object (Article 15, sections III and V). Using data for a purpose that was not disclosed requires obtaining consent again (Article 11).
What to do and by when
The right has been in force since March 21, 2025.
- Inventory the processes in which a system decides or makes recommendations about individuals: credit, collections, hiring, performance, personalized pricing and fraud prevention.
- Classify each process according to whether or not there is human intervention with real ability to modify the decision.
- For processes without human intervention, design a channel for review by a person that can be activated when the data subject objects.
- Document which processes respond to a legal obligation, with the legal basis that supports them.
- Update the privacy notice to disclose the automated evaluation purpose and the means to object.
- If the model is provided by a third party, make sure by contract that the vendor will allow you to suspend automated processing for a specific data subject.
Recommendations to mitigate risk
If the company fails to handle an objection, the authority may issue a formal warning ordering it to comply, and disregarding the warning or processing requests negligently is punishable by a fine of 100 to 160,000 Units of Measurement and Update, or UMA (MXN 11,731 to MXN 18,769,600), under Articles 58, sections I, II and VII, and 59, sections I and II. Continuing to use the data after the data subject or the Ministry of Anti-Corruption and Good Governance (Secretaría Anticorrupción y Buen Gobierno) requested that it cease, or processing the data in a way that prevents the exercise of these rights, is punishable by a fine of 200 to 320,000 UMA (MXN 23,462 to MXN 37,539,200), under Articles 58, sections XVI and XVII, and 59, section III. If the model processes sensitive data, such as health information, the penalties may be increased up to twofold (Article 59, section IV).
How we approach it at LEXURE
We map your company's automated processes, identify which ones fall under Article 26 and, together with your team, design the human review channel, the notice language and the contracts with model vendors. With that framework, your company can keep automating and has a clear procedure for handling a data subject who objects.
If your company makes decisions about individuals with the support of automated models, schedule a diagnostic meeting with us.
Legal basis
- Federal Law on the Protection of Personal Data Held by Private Parties, Articles 11, 15, sections III and V, 26, 31, 58, sections I, II, VII, XVI and XVII, and 59. Current text (opens in a new tab)
- 2026 value of the Unit of Measurement and Update, National Institute of Statistics and Geography (INEGI). Official publication (opens in a new tab)
This note is for information only and describes the law in force on the verification date. It is not legal advice on a specific matter. It is a translation of the Spanish original: Mexican law is published in Spanish, and the official Spanish texts prevail.