AI Usage Policy

Last updated: January 2, 2026

This English translation is provided for convenience only. In the event of any discrepancy or conflict between versions, the Portuguese version prevails.

1. Introduction and Contractual Integration

This Artificial Intelligence Usage Policy ("AI Policy") supplements and forms an integral part of the Terms of Service and the Privacy Policy of Arkar Tecnologia Ltda. ("Arkar"). This Policy establishes the guidelines, limitations and responsibilities related to the use of the artificial intelligence ("AI") features made available through the Platform.

By using any AI feature of the Platform, you represent that you have read, understood and fully accepted this Policy, as well as the Terms of Service and the Privacy Policy.

2. Definitions: Input, Output and Content

2.1. Input

"Input" means any data, information, document, instruction, query or content provided by the User to the Platform's AI systems, including prompts, uploaded documents, fund data, questions and configuration parameters.

2.2. Output

"Output" means any result, response, analysis, report, recommendation, text, data or content generated by the Platform's AI systems in response to an Input.

2.3. Intellectual Property

The User retains all intellectual property rights over their Inputs and over the Outputs generated from their Inputs, subject to the terms of this Policy and of the Terms of Service. Arkar retains all intellectual property rights over the underlying technology, including algorithms, AI models, software, interfaces and the Platform's infrastructure.

2.4. User Content

The User is solely responsible for the content they provide as Input and for how they use the Outputs generated by the AI. Arkar is not responsible for the accuracy, completeness or adequacy of the Inputs provided by the User.

3. AI Features

The Platform provides several features based on artificial intelligence, including:

  • Document Analysis: Automated processing and extraction of information from regulatory, contractual and operational documents.
  • Fund, Performance and Risk Analysis: Automated assessment of performance indicators, risk metrics and investment fund comparisons.
  • Automated Valuation: Asset valuation and pricing models based on artificial intelligence.
  • Regulatory Reporting: Automated generation of reports required by regulatory bodies (the Brazilian regulators CVM and ANBIMA).
  • Security and Monitoring: Detection of anomalies, suspicious activity and potential security violations.
  • Market Intelligence: Analysis of market trends, news and events relevant to the financial sector.
  • Automated Compliance: Automatic verification of compliance with regulatory rules and internal policies.
  • Chatbots and Agents (Arkar Fund Chat): Intelligent virtual assistants for queries, support and task automation.

4. Training and Use of Data

4.1. Anonymized data to improve models

Arkar may use anonymized and aggregated data to improve its artificial intelligence models and the quality of the services provided. No data that allows the identification of individuals or organizations will be used for training without express authorization.

4.2. Opt-in/opt-out controls

The User may choose whether or not to participate in AI model improvement programs through the Platform settings. Opting out will not affect the availability or quality of the core AI services.

4.3. Sensitive data

Data classified as sensitive (as defined in Lei nº 13.709/2018 (LGPD – Brazilian General Data Protection Law)) receives differentiated treatment and is not used for AI model training, except with the specific and substantiated consent of the data subject.

5. Output Limitations and Risks

Outputs generated by the AI are subject to limitations inherent to the technology and may exhibit:

  • Errors: AI models may generate incorrect, inaccurate or incomplete information.
  • Omissions: Outputs may not cover all relevant aspects of an analysis or decision.
  • Biases: Models may reflect biases present in the training data or in the methodologies used.
  • Failure to reflect actual facts: Outputs are generated on the basis of statistical patterns and may not correspond to factual reality.
  • Statistical estimates: Values, projections and metrics generated by the AI are estimates and must not be treated as definitive figures.
  • Incorrect interpretations: The AI may misinterpret the context, intent or meaning of the Inputs provided.

6. Mandatory Human Review

No Output should be used without appropriate human validation.

Every Output generated by the AI must be reviewed, validated and approved by a qualified professional before being used for any purpose, including investment decisions, regulatory reports, communications to investors and any other actions that may have legal, financial or operational impact.

7. User Responsibilities

The User is responsible for:

  • Reviewing and validating all Outputs before using them for any purpose.
  • Ensuring that the use of the AI complies with applicable laws and regulations.
  • Not using the AI for illegal, unethical or harmful purposes.
  • Immediately reporting to Arkar any Outputs that contain clearly incorrect, harmful or inappropriate information.
  • Keeping adequate records of the reviews and approvals of Outputs used in regulatory or investment processes.

8. Third-Party Technologies

The Platform uses third-party artificial intelligence technologies, including models developed by OpenAI and Anthropic. The use of these technologies is subject to the terms and conditions of the respective providers, in addition to the terms of this Policy. Arkar carefully selects and monitors its technology partners to ensure compliance with security, privacy and quality standards.

9. AI Security

Arkar implements specific security measures to protect its AI systems against:

  • Adversarial attacks: Attempts to manipulate the behavior of the AI models through malicious Inputs.
  • Model inversion: Attempts to extract training data from the models.
  • Data poisoning: Attempts to corrupt the data used to train or fine-tune the models.
  • Training inference:Attempts to deduce confidential information from the model's Outputs.

10. Prohibited Uses

It is expressly prohibited to use the Platform's AI features to:

  • Generate false or misleading information with intent to commit fraud or manipulate the market.
  • Reverse engineer the AI models or attempt to extract training data.
  • Use Outputs as the sole basis for investment decisions without human review.
  • Infringe third-party intellectual property rights through the AI.
  • Attempt to circumvent security measures, content filters or model limitations.
  • Process third-party personal data without an adequate legal basis.
  • Generate content that is discriminatory, offensive, illegal or harmful to third parties.

11. Disclaimers

THE AI SERVICES ARE PROVIDED "AS IS" AND "AS AVAILABLE". ARKAR DOES NOT WARRANT THAT THE OUTPUTS WILL BE ACCURATE, COMPLETE, ADEQUATE, TIMELY OR ERROR-FREE. ARKAR SHALL NOT BE LIABLE FOR ANY DECISIONS MADE ON THE BASIS OF AI OUTPUTS WITHOUT PROPER HUMAN VALIDATION.

THE USER ACKNOWLEDGES AND ACCEPTS THAT ARTIFICIAL INTELLIGENCE IS AN EVOLVING TECHNOLOGY AND THAT ITS RESULTS MAY BE UNPREDICTABLE, INCOMPLETE OR INCORRECT. THE USE OF THE OUTPUTS IS THE SOLE RESPONSIBILITY OF THE USER.

12. Limitation of Liability

Arkar's total liability in connection with the use of the AI features shall be limited to the total amount paid by the User in the 12 (twelve) months preceding the event giving rise to the claim. This limitation applies to the maximum extent permitted by applicable law.

13. Updates and Modifications

Arkar reserves the right to update this Policy at any time to reflect changes in the AI features, in the technologies used, in applicable law or in market best practices. Significant changes will be communicated with reasonable advance notice through the Platform or by email.

14. Educational Content Attribution

The artificial intelligence practices and methodologies adopted by Arkar have been influenced by several sources, including, as a reference, the AI Agent Creation Course (Curso de Criação de Agentes de IA), Protocolo Genesis – Parts 1 and 2, offered by the TOPHAWKS community and developed by Ronnald Hawk.

15. Contact

If you have questions, suggestions or complaints about this AI Usage Policy, please contact us: