Internal Policy for Use of Large Language Models (LLMs)

This policy outlines the guidelines for the appropriate and effective use of LLMs and LLM-enhanced software features within our organization. It applies to all employees, contractors, consultants, temporary staff, and other workers at the company, including all personnel affiliated with third parties. It covers all use of LLMs and LLM-enhanced software features in activities such as client communications, report creation, product writing, design, and any other business-related tasks. The purpose of this specific policy is to:

• Ensure the responsible and ethical use of LLMs in all business activities.
• Protect client confidentiality and company proprietary information.
• Maintain compliance with applicable laws, regulations, and industry standards.
• Enhance efficiency while mitigating risks associated with the use of AI technologies.

Definitions

• Large Language Models (LLMs): Advanced AI systems capable of understanding and generating human-like text based on deep learning algorithms.
• LLM-Enhanced Software Features: Software functionalities that incorporate LLMs to augment user capabilities, such as predictive text, content generation, or automated summarization.

General Guidelines

• Authorization: Only use company-approved LLM tools and software. Unauthorized use of external LLM services is prohibited unless explicitly permitted for specific, non-sensitive tasks under clear guidelines.
• Training: Users must complete requisite company training on the responsible use of LLMs and this policy before utilizing them in their work.
• Accountability: Users are ultimately responsible for the output generated or assisted by LLMs they use. This includes verifying accuracy, ensuring appropriateness, and checking for potential issues like bias or plagiarism before using the output.
• Transparency: Disclose when content is generated or significantly assisted by an LLM where appropriate or required by company guidelines, particularly in external communications or formal reports.

Use in Client Communications

• Accuracy and Reliability: ALWAYS verify all LLM-generated communications for factual accuracy, contextual appropriateness, and tone before sending to clients. Do not rely solely on AI output.
• Confidentiality: STRICTLY PROHIBITED: Do not input, paste, or otherwise disclose confidential client information or personal data into public or unauthorized LLMs. Only use approved, secure tools where data handling complies with privacy laws (e.g., GDPR), client agreements, and company policies.
• Professionalism: Ensure all AI-assisted communications strictly adhere to the company’s standards for professionalism, brand voice, and tone.

Use in Report Creation

• Quality Assurance: Thoroughly review and critically edit LLM-generated reports or sections to ensure they meet the company’s quality, accuracy, and analytical standards. Add necessary context, analysis, and verification.
• Attribution: If required by company guidelines or academic/professional standards, attribute the use of LLMs in the creation of reports appropriately.
• Data Security: Avoid inputting sensitive company financial data, strategic plans, or other proprietary information into LLMs that lack explicit company approval and appropriate security assurances.

Use in Product Writing and Design

• Innovation Support: LLMs can be leveraged for brainstorming, idea generation, and drafting initial concepts. However, all final product descriptions, designs, or creative outputs must be reviewed by humans for originality, accuracy, compliance, and strategic alignment.
• Compliance Checks: Ensure all AI-assisted content complies with intellectual property laws (copyright, trademark) and does not infringe on third-party rights. Verify originality where needed.
• Ethical Considerations: Avoid using LLMs to create content that is knowingly false, misleading, biased, discriminatory, harmful, or otherwise unethical.

Data Privacy and Security

• Personal Data Protection: Do not input personal data (of clients, employees, or others) into LLMs unless explicit consent has been obtained where required, data processing agreements are in place (if applicable), and the tool used is company-approved for handling such data under GDPR or other relevant regulations.
• Data Minimization: When using approved tools for tasks involving necessary data, only provide the minimum amount of information required for the LLM to perform the task effectively.
• Secure Usage: Ensure that any interaction with LLMs, especially if involving company data, occurs through secure connections and approved platforms. Follow company guidelines for storing or handling outputs that may contain sensitive information.

Compliance and Legal Considerations

• Regulatory Adherence: All use of LLMs must comply with applicable laws and regulations, including data protection laws (GDPR, CCPA, etc.) and any industry-specific regulations relevant to our business.
• Intellectual Property Rights: Respect all copyrights, trademarks, patents, and other intellectual property rights. Do not use LLMs to generate content that infringes on these rights. Understand the ownership and usage rights of AI-generated output based on the tool’s terms.
• Third-Party Policies: When using company-approved third-party LLM services (e.g., via API), adhere strictly to their terms of service and usage policies.

Training and Support

• Mandatory Training: All relevant personnel must participate in mandatory training sessions covering this policy, ethical AI use, data security practices, and effective prompting techniques for approved tools.
• Resource Access: Utilize company-provided resources, guidelines, and designated support channels for assistance or clarification on LLM-related tasks and policies.
• Continued Education: Stay informed about updates to this policy, approved tools, and emerging best practices regarding LLM usage in a business context.

Monitoring and Review

• Usage Monitoring: Be aware that the company may monitor the usage of approved LLM tools to ensure compliance with this policy and manage associated costs or risks.
• Feedback Mechanisms: Promptly report any issues, unexpected outputs, potential security concerns, or ethical dilemmas related to LLM use to your supervisor or the designated compliance officer/department.
• Policy Review: This policy will be reviewed periodically (e.g., annually or as needed) and updated to reflect technological advancements, evolving regulations, and business needs.

Responsibilities

• Employees/Users: Adhere strictly to this policy, complete required training, use LLMs responsibly and ethically, ensure the accuracy and appropriateness of AI-assisted work, and report concerns.
• Managers: Ensure team members are aware of, understand, and comply with this policy. Provide necessary support and guidance regarding the appropriate use of approved LLM tools.
• Compliance Officer/Designated Body: Oversee adherence to this policy, manage approvals for LLM tools (if applicable), handle reported violations or concerns, and coordinate policy reviews and updates.

Violations and Disciplinary Actions

• Non-Compliance: Failure to comply with this policy may lead to disciplinary action, which could range from retraining or warnings up to and including termination of employment or contract, depending on the severity of the violation.
• Reporting Violations: Employees are encouraged and expected to report any suspected violations of this policy through appropriate channels (e.g., manager, compliance department, anonymous hotline if available).

Acknowledgement

• All employees and relevant personnel are required to formally acknowledge that they have read, understood, and agree to comply with this internal policy regarding the use of Large Language Models.