Publications

Internal Policy on the Use of Artificial Intelligence: Why Companies Should Establish Rules Before Facing an Incident

October 30, 2026

Internal Policy on the Use of Artificial Intelligence

 

Artificial Intelligence in the Corporate Environment and the Governance Challenges

Artificial intelligence is transforming the daily operations of companies, enabling them to optimize processes, accelerate analyses, and increase productivity. However, its use also brings legal and operational challenges related to data protection, information confidentiality, intellectual property, and labor relations. In this scenario, the absence of internal guidelines can expose organizations to risks that could be prevented through appropriate governance measures.

Imagine an employee who needs to urgently review a contract and, to expedite the task, uses a free artificial intelligence tool, entering the entire document, including names, values, and business information. Although the response is produced in seconds, the company may be unaware of the storage and usage conditions of this data, as well as the risks arising from sharing it with an external platform.

 

The Phenomenon of Shadow AI and the Importance of Domestic Policy

Situations like this illustrate the phenomenon known as Shadow AI, This is characterized by the use of artificial intelligence tools without the organization's authorization, supervision, or governance. The problem is not necessarily the technology itself, but the absence of criteria for its use, especially when corporate information is shared without prior assessment of security conditions and contractual terms.

Developing an internal artificial intelligence policy is therefore a measure of corporate governance and risk prevention. Its objective is to establish responsibilities, define authorized uses, guide employees, and reconcile technological innovation with the company's legal obligations.

 

Personal Data Protection (LGPD) and Liability of Data Agents

One of the main points of attention involves the protection of personal data. The use of artificial intelligence tools does not preclude the application of the General Data Protection Law (Law No. 13.709/2018), which establishes obligations related to purpose, adequacy, necessity, security, and prevention of harm. The insertion of personal data on external platforms must observe an appropriate legal basis and other requirements applicable to data processing.

Responsibility is not automatically transferred to the tool provider. Its definition depends on the activities performed by each agent and the specific circumstances. Contracting third parties does not exempt the company from assessing the risks involved, adopting security measures, and verifying the conditions for handling information.

Failure to comply with the LGPD (Brazilian General Data Protection Law) may result in administrative sanctions applied by the National Data Protection Authority (ANPD), including a fine of up to 21 times the revenue of the private legal entity, group, or conglomerate in Brazil, excluding taxes, limited to R$ 50 million per infraction. Other sanctions provided for in Article 52 of the law may also apply, without prejudice to any civil liabilities, as the case may be.

 

Protection of Trade Secrets and Contractual Safeguards

In addition to personal data, the company must protect confidential information such as contracts, business strategies, corporate documents, financial data, and trade secrets. Sharing this information with external tools without assessing security controls and contractual conditions can compromise business interests and lead to significant legal consequences.

Therefore, before authorizing a tool, it is necessary to verify whether the data entered is used for model training, how long it is stored, what security measures are adopted, and under what conditions it can be shared with third parties. When there is an international transfer of personal data, articles 33 to 36 of the LGPD (Brazilian General Data Protection Law) and applicable regulations must also be observed.

 

Governance in Automated Decision-Making Processes and Human Review

The governance of artificial intelligence must also encompass decision-making processes. Tools used in credit analysis, candidate selection, supplier evaluation, and the definition of commercial terms can produce inadequate or discriminatory results. In this context, Article 20 of the LGPD (Brazilian General Data Protection Law) ensures the data subject's right to request a review of decisions made solely based on automated processing of personal data that affect their interests. Article 6, IX, in turn, establishes the principle of non-discrimination, prohibiting the processing of data for illicit or abusive discriminatory purposes.

The adoption of human evaluation and review mechanisms, when relevant to the risks of the application, helps to identify inconsistencies and correct inadequate results. Recording decisions, those responsible for validation, and the procedures adopted also promotes traceability and demonstrates the governance measures implemented.

 

Implications for Intellectual Property and Labor Relations

Intellectual property represents another relevant aspect. The use of artificial intelligence in the creation of texts, images, codes, and other materials requires attention to the ownership of rights and the conditions for using the results. In Brazil, copyright protection depends on the requirements stipulated in Law No. 9.610/1998, and must be analyzed according to the circumstances of the creation and the human creative contribution. Furthermore, the generated content may present similarities to pre-existing works or reproduce protected elements, making its verification advisable before commercial use or publication.

Workplace relations also deserve attention. Misuse of tools can involve the sharing of confidential information or non-compliance with corporate rules. The existence of a formally communicated policy, accompanied by training and proof of employee awareness, helps to establish clear expectations and guide the adoption of disciplinary measures. Reviewing employment contracts, confidentiality agreements, and codes of conduct may also be necessary. The validity of any sanction, however, will depend on the specific circumstances and the requirements of labor law.

 

Risk Management, Regulatory Landscape and Practical Measures

The adoption of artificial intelligence tools should be integrated into the company's risk management. In addition to the terms of use, security mechanisms, contractual responsibilities, data storage and deletion conditions, and audit possibilities should be evaluated. Depending on the nature of the operation, it is recommended to negotiate specific clauses regarding confidentiality, data protection, intellectual property, and incident reporting.

In terms of regulation, Bill No. 2,338/2023, intended to establish the legal framework for artificial intelligence in Brazil, was approved by the Federal Senate in December 2024 and sent to the Chamber of Deputies in March 2025. The legislative process should be monitored to verify any changes to its status and the proposed text. Regardless of the approval of specific legislation, norms such as the LGPD (Brazilian General Data Protection Law), the Civil Code, the Consolidation of Labor Laws, and copyright law already offer applicable legal parameters for the corporate use of technology.

An internal artificial intelligence policy should translate these parameters into objective procedures. Key measures include classifying authorized, restricted, and prohibited tools; defining information that cannot be entered into external platforms; indicating uses that require human review; identifying those responsible for approving new solutions; and establishing procedures for reporting and investigating incidents.

 

Effectiveness of Policy and Preventive Legal Action

It is also recommended to implement periodic training, access controls, and supplier evaluation criteria, as well as to review the policy according to technological, contractual, and regulatory changes. Its effectiveness depends not only on the drafting of the document but also on its practical application, the assignment of responsibilities, and the ability to verify compliance with the rules.

In this context, internal artificial intelligence policy should be understood as part of the corporate governance, risk management, and compliance framework. When properly implemented, it allows for the reduction of unnecessary exposures, the protection of strategic information, and greater predictability in the use of new technologies, without hindering innovation.

Preventive legal action is fundamental to this process. Specialized counsel can assist in identifying risks, developing internal policies, reviewing contracts with technology providers, and adapting confidentiality and data protection instruments. In this way, artificial intelligence can be incorporated into business activities in a structured manner, reconciling operational efficiency, innovation, and legal security.

Our legal team monitors legislative and regulatory discussions related to artificial intelligence and is available to advise companies on the development and review of internal policies, the assessment of legal risks, and the adaptation of contracts and corporate procedures.

 


Article written by: Leonardo Neri, Isabela Melo and Emilly Fukuoka.

  • “"Artificial intelligence should not be treated merely as a productivity tool, but as part of the company's governance and risk management framework."”
  • “"The absence of an internal policy does not eliminate the organization's legal obligations. Prevention begins with defining clear rules, responsibilities, and controls that are compatible with the risks involved."”
  • “An effective policy doesn’t need to stifle innovation. Its role is to establish safe boundaries so that artificial intelligence is used responsibly, in line with legal obligations and business interests.”
  • “"The evaluation of an artificial intelligence tool must go beyond its functionalities and cost. The conditions for data processing, the confidentiality of information, and contractual responsibilities also need to be factored into the decision."”
  • “"The governance of artificial intelligence is not simply about drafting a document. Its effectiveness depends on the practical application of the rules, the training of employees, and the periodic review of the controls adopted."”

If you have any questions about the topics covered in this publication, please contact any of the lawyers listed below or your usual Mazzucco&Mello contact.

Vitor Antony Ferrari

+55 11 3090-9195

Ivan Kubala

+55 11 3090-9195

Antonio Carlos Cantisani Mazzucco

+55 11 3090-9195

Leonardo Neri Candido de Azevedo

+55 11 3090-9195

Rafael Mello

+55 11 3090-9195

This communication, which we believe may be of interest to our customers and friends of the company, is intended for general information only. It is not a complete analysis of the matters presented and should not be considered legal advice. In some jurisdictions, this may be considered lawyer advertising. Please see the company's privacy notice for more details.

Related Areas

Related Professionals