Generative “AI” Use Policy

This policy governs the use of generative “artificial intelligence” (AI) tools or platforms (“GenAI”) by any person in the performance of services for or on behalf of The Minerva Group.

Any questions about this policy should be directed to Joseph Councilman.

What is Generative Al?

Generative artificial intelligence refers to technology that creates new content (such as text, audio, data, images, video) (“Output”) by leveraging content that the technology was trained on (e.g., through machine learning) (“Training Data”) in response to prompts submitted by a user (“Prompts”).

The following are examples of Generative Al platforms (“GenAI Platforms”): 

Type of GenAI	Examples
Text-to-text	CPA Pilot, Chat-GPT, Copilot, Gemini, Grammarly AI Writing
Text-to-images	DALL-E 2, Stable Diffusion, Midjourney
Text-to-source code	Chat-GPT, GitHub CoPilot, Open Al Codex
Text-to-video	Synthesia, Runway
Text-to-sound	Amper
Text-to-voice	UberDuck

Appropriate Use

Some appropriate business uses for Generative Al include:

• Sparking creativity
• Research
• Automating or assisting with repetitive tasks
• Generating content for social media and marketing
• Rapid messaging and correspondence
• Summarizing meetings or conversations
• Generating product and business ideas or strategies

Defining “Data” Types

There are four (4) basic types of data:

1. Public Data: This information is available to the public, for example, contents of a Wikipedia article.
2. Non-Confidential Data: In addition to public information, this can be non-public information that does not require special protection. It could include information you receive from a client that is not considered confidential or sensitive, or subject to access controls. Non-confidential data should not contain any sensitive information or any personally identifiable information. For example, general business data and anonymized data would be non-confidential. If you redact confidential information out of a data set, the remaining data may be non-confidential.
3. Confidential Information: This is client information that is covered by contractual confidentiality restrictions or information licensed from a third party.
4. Personal Information: This includes information such as names, addresses, phone numbers, SSNs, etc.

Here is a suggestion for how to categorize these data types with available GenAI tools.

Type of Data	Risk Level	Tools Available for Use
Public Data	Low	Generally safe to us in any GenAI tools
Non-Confidential Data	Low	Generally safe to us in any GenAI tools
Confidential Data	Moderate	Only firm-sanctioned tools that have data protections
Personal Information (PII)	High	Only firm-sanctioned tools that have data protections

GenAl Tool Selection

Until we, as an organization, better understand the capabilities and risks of the various emerging and generally available Al tools, and products, is it company policy that you limit your work-related interactions with GenAI Platforms. Listed below are the tools/platforms that have been evaluated, along with appropriate use guidelines.

List of Approved Apps by Data Types

Type of Data	Approved Apps	Use guidelines
Public Data	CPA Pilot, ChatGPT, Copilot & Gemini	Research, initial draft of communication, generate ideas to assist with tasks or summarize/analyze information.
Non-Confidential Data	CPA Pilot, ChatGPT, Copilot & Gemini	Research, initial draft of communication, generate ideas to assist with tasks or summarize/analyze information.
Confidential Data	CPA Pilot ONLY (with extreme caution)	If absolutely necessary, confidential data may be input, sent, uploaded or otherwise conveyed to CPA Pilot only, not other a GenAI system should be used.
Personal Information (PII)	CPA Pilot ONLY (with extreme caution)	If absolutely necessary, PPI data may be input, sent, uploaded or otherwise conveyed to CPA Pilot only, not other a GenAI system should be used.

Note: If you have GenAl needs that are not being met by these provided tools, please contact Joseph Councilman to discuss options and alternatives. Your specialized, unmet use case may provide an opportunity to participate in early-release, beta, pilot, or as-needed software trials.

Policy Enforcement

This policy is not optional. The Company takes use of GenAl seriously and has developed a team to be responsible for governance and oversight of the Al policy. If you are unclear how to proceed in a particular scenario, please reach out to Kris Byers or Joseph Councilman.

Failure to comply with this policy could result in disciplinary action, up to and including termination. If you become aware of any potential violations, please report these to Joseph Councilman as soon as you are made aware, as an information breach could result in significant loss to the company.

General Guidance

We must only use GenAI Platforms in ways that are lawful, transparent, and fair.

As with all technology it is good to trust what it helps you produce; however, you should always validate and verify the output.

Permitted Behavior

GenAI Platforms may be used in a manner that is not restricted or prohibited by this Policy, such as:

• a starting point for ideas or inspiration
• a replacement for a search engine and for information gathering
• a way to summarize or analyze non-confidential or personal information

Prohibited Behavior

The following uses of GenAI Platforms are prohibited:

• DO NOT use Prompts or Outputs that are harmful, inappropriate, or violate the Company’s standards and policies concerning appropriate workplace conduct.
• DO NOT use Outputs for any fact-based use cases where accuracy is important, without additional verification.
• The raw output of GenAl tools must never be used as the determining factor in decision making which may have material or legal effects upon natural persons, including The Minerva Group employees or clients.

Required Behavior

Any employee that uses a GenAI Platform for business purposes must adhere to the following requirements:

Accuracy

• DO review any Output to identify errors, security vulnerabilities, plagiarism and other infringement risks, and other issues.

Records

• DO keep records of Outputs that are intended for external use cases.
• DO keep records of any GenAI Platforms you use for business purposes.

Transparency

• DO NOT hide from, or mislead, any third party regarding the use of artificial intelligence in any Outputs.
• If using GenAI Platforms in work for a client ALWAYS confirm that such uses are permitted under the client relationship.

GenAI Platform Guidelines

• DO comply with any usage policies and other terms of the applicable GenAI Platform. For questions on any such policies, please contact Kris or Joseph.