Country Profile – Mexico

Mexico’s National Electoral Institute (Instituto Nacional Electoral; INE) has prioritized building institutional AI-readiness before implementing AI technologies. Through investments in upgraded hardware, enhanced training data for AI systems and strengthened cybersecurity protocols, INE is establishing a robust digital infrastructure. This foundation not only supports immediate improvements in efficiency but also positions the INE to leverage AI for broader, transformative advancements in electoral management.

At the same time, INE recognizes that technological upgrades must be accompanied by changes in institutional policy and culture, to ensure that technologies are meaningfully embraced by end-user practitioners and that they offer viable solutions to their problems. This entails the formation of dedicated AI-units, internal guidelines and AI literacy efforts that support staff across INE in harnessing the potential of AI as an election technology, while also ensuring that practitioners recognize risks and limitations. This institutional infrastructure includes standardized pathways for assessing, acquiring, developing and evaluating AI-solutions, which are grounded in institutional policy and regulatory compliance.

How is AI used to improve electoral management in Mexico?

The National Electoral Institute has already made targeted investments in integrating AI into its election administration processes. It is also continuously exploring new areas where AI can enhance the accessibility, structure and efficiency of INE operations across a wide range of applications.

INE’s current use of AI can be grouped into three main areas:

• Legal research and public inquiry response: AI-tools are being implemented to support the processing of public inquiries by helping legal departments at INE to categorize and draft preliminary responses to incoming letters. As INE typically receives as many as two thousand letters per day, the AI pre-response tool helps practitioners efficiently categorize queries and prepare the foundational elements of formal replies. To ensure output accuracy, models are trained on relevant electoral policy and regulatory data, and all outputs are reviewed by the legal department before responses are finalized and transmitted.
• Social media campaigns: AI is used in broader public communications and voter engagement, for example to generate images used in social media content posted by INE’s official channels. The use of generative AI in INE’s content sparked some controversy in 2025, when the AI-generated voiceover in a video posted to TikTok closely resembled the voice of a recently deceased local musician. This incident was a key motivator for INE to develop standardized operating procedures governing the use of AI, including an internal authorization process for specific tools, models and applications.
• Cybersecurity reinforcement: At the end of 2025, the Federal Government of Mexico issued a unified National Cybersecurity Plan with the intention of standardizing procedures for threat mitigation and response across all levels of government (Gobierno de Mexico 2025). While INE is aligned with this new mandate, it has already surpassed several of the Plan’s benchmarks for technical safeguards, incident-reporting mechanisms and staff training. Building on this foundation, INE has begun integrating AI-tools into its cybersecurity frameworks, including third-party AI agents for detecting threats and vulnerabilities, in order to enhance the speed and precision of defensive responses.

Areas where AI tools are currently under consideration

The National Electoral Institute is also exploring a range of potential areas where AI can improve existing structures or introduce innovative practices, while carefully assessing the safeguards and capacity requirements that must be in place before deployment. This approach ensures that any technological adoption complies with existing regulation and does not undermine voters’ equitable access to their democratic rights.

The two main examples of concrete AI-related initiatives that INE is currently pursuing are the integration of large language models (LLMs) into INE’s virtual assistant for voters and the broader ‘data lake’ initiative:

• Multilingual voter information chatbot: For several years, INE has hosted a chatbot under the name ‘INÉS’ to supply voters with basic information about Mexican elections and the work of the Institute. INÉS currently has no AI-based components, instead functioning as an interactive voice response (IVR) that provides preset answers to simpler questions, and which redirects users with more complex inquiries to the appropriate communication channel. INE recognizes the potential improvements that could be made by integrating deep learning technologies to expand INÉS’s capabilities, particularly with multilingual support for the near three hundred languages spoken across Mexico. However, significant concerns remain regarding the risk of AI misconstructions—‘hallucinations’—particularly as large language models tend to exhibit higher error rates in lesser-spoken languages due to their limited prioritization in training data. To address this issue, INE is actively reviewing how issues relating to accuracy and reliability can be addressed definitively before integrating the technology into the existing INÉS system.
• The ‘data lake’ initiative: INE’s data lake strategy seeks to combine (1) existing voter and election data stored by the Institute with (2) the shared data resources from agreements with other public institutions and (3) publicly available data banks, to craft a flexible, schema-on-read data pool upon which AI systems can grow. INE’s Government Group on Information and Communication Technology (GGTIC) intends to employ the data lake for operational modelling to, for example, determine polling station placements and resource allocation, or for the financial auditing and oversight of political parties. In the long term, INE plans on integrating the data lake with machine-learning technologies to design systems that can support the Institute beyond the capabilities of existing statistical modelling. To achieve this, INE will utilize a data-driven approach, where correlations between different data points are utilized to explore new meaningful applications for AI technologies.

Crucially, the data lake strategy is reliant on cross-institutional interoperability of systems and data. INE is partnering with other state bodies in inter-agency data-sharing arrangements, which are essential to satisfy the required levels of data-set diversity and saturation. The data lake is still under development, and systems will undergo rigorous piloting and assessment to guarantee that they meet the internally set benchmarks for validity and reliability, created with the high-risk sensitivities of elections in mind.

The National Electoral Institute’s guidelines on the Use of AI

In August 2025, INE became the first government authority in Mexico to ratify a nationwide regulatory policy on the responsible use of AI technologies. Two key instruments—the ‘Guidelines and Principles for the Strategic Development and Regulated Use of Artificial Intelligence’ and the ‘Institutional Work Programme 2025–2026’—provide institutional directives on authorized AI use and address issues of compliance with regulatory, ethical and democratic norms. The ‘Guidelines and Principles’ specifically acknowledge that AI should act as a complement to human practitioners, and that AI system development at INE follows a human-centric design philosophy. This includes the involvement of human oversight in system management and evaluation, particularly where the outcomes of decision-making processes affect constituents’ abilities to exercise their democratic rights and liberties.

The ‘Guidelines and Principles’ also recognize that the trustworthy, confident and cohesive implementation of AI can only be secured by nurturing institutional capacity and organizational culture. To this end, INE has committed to conducting internal digital proficiency training programmes that encompass AI literacy and cyber hygiene at all staff levels. As of January 2026, INE has introduced an institution-wide effort to acquaint staff members with the technical fundamentals of AI, including its limitations and risks, as well as its legal, ethical and data-protection implications. This programme provides staff with the skills to review AI-generated results thoughtfully, ensuring decisions are made under careful human supervision. Through this process, team members can actively uphold the institution’s values, as outlined in the ‘Guidelines and Principles’, while recognizing potential discriminatory biases or risks of error.

The National Electoral Institute’s AI-unit: building structural support for electoral AI 

INE’s investments in AI technologies are accompanied by an ecosystem of managerial units with the mandate to make decisions on when and how AI should be used, as well as of the necessary precautions for each application. This includes the founding of a dedicated AI unit, which ensures that each AI application adheres to governance principles, including—but not limited to—algorithmic transparency, equity and non-discrimination, traceability and accountability, data protection and privacy, algorithmic impartiality and certainty, and digital legality. In practice, the AI unit supports other departments of the Institute in advising on the propriety of AI in different processes, based on its expert judgement rooted in the ‘Guidelines and Principles’. The centralized, overarching nature of this policy allows for a unified adoption of AI-tools across all branches of the Institute, with sufficient concrete instructions to make the implementation of AI clear and feasible.

In addition to the AI unit, INE also has a dedicated technology committee for the purpose of evaluating technological acquisitions. The committee is led by INE’s Executive Secretary and is comprised of personnel from the GGTIC and key representatives from other directorates whose substantive functions involve the use of election technologies. These include, among others, departments managing voter registration, political party oversight and administrative operations. Evaluation procedures are based on collegiate decision making that balance operational needs, priorities and concerns across the different branches of INE in line with the institution-wide AI policy.

The development or acquisition of AI systems is subject to a formal, multi-stage validation process. Once a potential use-case is identified, the unit in question decides whether it seeks to develop an in-house AI solution or acquire an existing third-party service. Upon making the decision, the unit must request a technical opinion from the GGTIC, submitting documentation on the proposed system’s functionality, risks, data requirements and compliance with institutional principles. Thereafter, the GGTIC conducts an initial review of the submission before an approved application can advance to a formal evaluation process conducted by the acquisition committee.

Outside the acquisition process, end-users at INE are authorized to use only the AI-tools that have been previously approved; and individual uses of existing institution-wide tools must adhere to terms specified by INE’s Guidelines & Principles on AI. The formal appraisal process for acquiring and utilizing externally developed AI tools ensures that those systems meet the same standards of transparency, security, traceability and human oversight as internally developed tools, thereby mitigating risks relating to vendor dependence, opaque algorithms and the misuse of data. By institutionalizing the review process for developing and introducing new AI-tools, INE strengthens its capacity to preserve technological autonomy while keeping a door open to the responsible integration of third-party tools.