AI Using Generative AI for Advocacy: Risks and Benefits Summary Generative artificial intelligence (AI) refers to systems that can create human-like text, images, and other content in response to prompts. This document will explain how to responsibly use tools like ChatGPT, Claude, Gemini, Copilot, and Meta AI to assist with writing, research, social media posts, strategy development, and campaign planning. Generative AI is a potentially transformative technology that can dramatically increase efficiency and expand capacity. At the same time, the hazards are clear: Loose privacy standards, enormous energy needs, fake – but believable – content creation, “hallucinations,” and more make AI a challenging technology to use ethically. However, AI is here to stay, and it is quickly inserting itself into every aspect of technological life. To keep up with your adversaries and allies, it is necessary to learn to use AI responsibly while implementing strong guardrails for privacy protection and establishing strict organizational policies around data sharing, content quality, and human oversight. Benefits of Generative AI for Advocacy Content Creation and Communication Generative AI excels at quick content development, enabling users to quickly create draft content including letters to policymakers, social media posts, talking points for spokespeople, internal memos and, depending on the platform, images. While not without risks, using AI to create drafts of such content can be a reasonable starting point for small organizations or groups, especially those that lack staff, time, or financial resources. Analysis from Multiple Sources AI tools synthesize information from multiple source documents to create comprehensive outlines and work plans. Some platforms allow users to create folders or projects that store source documents, creating a knowledgebase to draw from when creating outlines or draft content. This capability is particularly valuable when incorporating research from various reports, studies, and policy documents; or when writing complex policy documents that might otherwise take hours to analyze. This work to synthesize and write content can be assisted by AI but should not be accomplished by AI, the latter of which risks publishing inaccurate or untrusted information in an inauthentic voice. Audience-Specific Messaging Organizations can use AI to quickly draft tailored messaging for different audiences and demographics using templates. This allows for more personalized communication without requiring extensive additional staff time for customization – though it is of course essential for human staff to edit and review any content drafted by AI. How AI Projects and Knowledge Bases Work About “Projects” AI tools don't remember previous conversations, so each interaction starts from scratch. Projects or folders allow you to create a persistent knowledge base that the AI can reference across multiple conversations. When you upload documents to a project, the AI can draw connections between different sources and provide more comprehensive, contextually-aware responses. How AI "Learns" from Your Documents AI doesn't actually learn or update from your uploads - instead, it temporarily incorporates your documents into its working memory for that session. Think of it like giving the AI a stack of reference materials to consult while answering questions. The AI will search through your uploaded documents to find relevant information and combine it with its general training knowledge. AI tools can only work with what you explicitly provide. If you upload incomplete information or documents that contradict each other, the AI may give you incomplete or conflicting responses. It won't know what you haven't told it. AI Tools Are Experimental and Error-Prone These tools are new, rapidly evolving, and make frequent mistakes. They can: Confidently state false information that sounds plausible Miss key details in documents you've uploaded Combine information incorrectly from multiple sources Generate outdated information based on their training data Fail to understand nuanced political or cultural contexts The AI Doesn't Always Answer Your Actual Question AI tools often respond to what they think you're asking rather than what you actually asked. They may: Focus on the wrong aspect of a complex question Provide generic advice when you need specific guidance Miss the strategic or political implications of your situation Give you technically correct but practically useless information Always Verify and Cross-Check Fact-check all claims, statistics, and citations Test AI recommendations on a small scale before full implementation Consult human experts for strategic decisions Review AI output with fresh eyes - does it actually address your needs? How to Communicate Effectively with AI Tools Be Specific and Direct Instead of: "Help with our campaign" Try: "Draft three key talking points about our opposition to the proposed energy project, focusing on environmental impact, using a tone appropriate for voters in key districts" Provide Context Explain who your audience is Specify the format you need (email, social post, policy brief) Include relevant constraints (word count, deadline, tone) Mention what you've already tried or what hasn't worked Use Follow-Up Questions "Can you make this more specific to [your location/issue]?" "This seems too generic - can you make it more compelling?" "What evidence would make this argument stronger?" Ask for Citations and Sources Always include: "Please cite your sources and provide links where possible" or "What evidence supports this recommendation?" Iterate and Refine Don't expect perfect results on the first try. Use responses like: "This is close, but can you focus more on [specific aspect]?" "The tone isn't quite right - can you make it more [urgent/professional/accessible]?" "Can you provide three different approaches to this same message?" Test Your Prompts Ask the same question in different ways to see if you get consistent, useful responses. If results vary wildly, your question may need to be more specific. Use Cases Building Knowledge Bases with Projects/Folders Setting Up Organizational Knowledge In the paid versions of ChatGPT or Claude, you can create “Projects” for each campaign, issue, or project you are working on, where you can upload background research, policy documents, approved messaging, and other resources. These resources act as a memory and knowledge base for your work. Upload public documents only: published reports, public statements, press releases, and general research materials Use these knowledge bases to generate consistent talking points, background summaries, project outlines, and educational content Example: A folder containing public reports, published studies, your organization's past public statements, and other non-sensitive materials can help generate draft content and outlines Best Practices Rule of thumb: Only upload materials you'd be comfortable making public Regularly audit folder contents to remove outdated information Create separate projects for different spheres of work (policy work, communications, organizational strategy) Safe Data Practices - Working Without Sensitive Information Document Sanitization Workflow: Create "clean" versions of documents with sensitive information removed (names, addresses, internal strategy details, donor information) Use placeholder text like "[ORGANIZATION NAME]" or "[LOCAL REPRESENTATIVE]" that you can customize after AI generates content When possible, work with publicly available data and research rather than internal documents, and only use internal documents that you’d be comfortable making public Template-Based Approach Develop messaging templates that AI can customize for different contexts Example: "Create three versions of this message - one for suburban voters, one for rural communities, and one for urban areas" Use AI to adapt public-facing content rather than create from scratch Research and Analysis Multi-Source Synthesis: Upload multiple public documents on the same issue and ask AI to summarize, identify common themes, highlight conflicting viewpoints, or note gaps in coverage Generate literature reviews from publicly available academic papers and policy reports Create comparative analyses of policy papers using publicly available documents Opposition Research: Analyze publicly available voting records, statements, and policy positions Generate fact-check summaries of public claims by opponents Always verify AI findings with original sources Content Adaptation and Translation Audience-Specific Messaging: Take existing public content and adapt it for different platforms (social media posts, newsletters, policy briefs) Adjust reading level and tone for different audiences Generate multiple versions for A/B testing Multilingual Outreach: Translate public-facing materials (with human verification) Adapt cultural messaging for different communities Create culturally appropriate versions of the same core message Risks and Challenges Data and Privacy Concerns Sensitive Information Exposure Organizations must assume that any information shared with AI tools will become public. The most critical risk involves uploading sensitive data. Names, addresses, contact information, and other personally identifiable information could potentially be exposed through platform tracking of uploaded data or via data breaches, or it could be incorporated into AI training datasets. Which means this information can be used inappropriately by anyone using the AI platform. Strategic Intelligence Leaks Campaign strategies, lists of names or email addresses, internal planning documents, and sensitive details about advocacy efforts should never be uploaded to AI platforms. The ways in which AI platforms digest, use, and share uploaded data are opaque at best, meaning there’s no clarity on whether or how uploaded data maybe be displayed to other users. Competitors or opposition groups could potentially gain access to this information, compromising campaign effectiveness. Legal Vulnerabilities AI-generated content may infringe on copyright, create misleading representations, or expose organizations to legal liability. As always, humans **must **review and edit such content before sharing it publicly. The legal landscape around AI-generated content remains unsettled, creating additional uncertainty. Content Quality and Authenticity Issues Generic "AI-Speak" AI tools, attempting to mimic authentic speech, can often end up using patterns and phrasing that can feel impersonal or inauthentic. Organizations must be sure to customize AI-generated drafts to ensure their work is in their own authentic voice. Factual Inaccuracies AI systems can confidently present false information, generate non-existent statistics, or misrepresent facts—a phenomenon known as "hallucination." All AI-generated content requires careful fact-checking and verification. Bias Amplification AI systems can perpetuate and amplify harmful stereotypes present in their training data, subtly inserting biased language or assumptions into messaging that could contradict organizational values and may go unnoticed without careful review. Loss of Authentic Voices Too much reliance on AI-generated content risks replacing genuine voices with artificial alternatives, potentially undermining the authenticity that gives your work its moral authority and emotional resonance. Limitations with Numerical Data and Calculations While AI tools can be helpful for analyzing trends and patterns in data, they have significant limitations when working with precise numerical calculations and datasets.  AI systems may introduce errors in mathematical computations, misinterpret numerical relationships, or provide results that appear accurate but contain subtle mistakes. Users should always verify numerical outputs through independent calculations or specialized analytical software, especially for data that will inform important decisions. When using AI for data analysis, treat the results as a starting point for further verification rather than as definitive findings. For critical numerical work, consider using dedicated statistical software or spreadsheet applications alongside AI tools rather than relying on AI alone. Operational and Strategic Risks Diminished Human Judgment Too much dependence on AI tools can lead to a reduced focus on critical thinking and strategic analysis; it’s easy to just let the robots do the work if you’re not careful! Empathy, critical thought, and relationship-building remain essential to effective advocacy and there are no replacements for them. Financial Considerations Premium AI subscriptions – often necessary for anything beyond one-off questions – can incur significant ongoing costs, which may be particularly challenging for smaller organizations to bear. Best Practices for Safe AI Adoption Any AI-created materials must undergo a process of human review before publication! Research To avoid an AI tool sharing false information with you while doing research, in your prompt,  always tell it to cite its findings for you. Verify that all links it shares are accurate before using the citation. For example, make sure the link the AI references actually works and leads you to where the AI found the research. Data Protection Protocols Establish clear guidelines for what information can and cannot be shared with AI tools. Create workflows that strip sensitive data before uploading and maintain secure processes for handling confidential information. Human Oversight Requirements Ensure that a human reviews – and edits, if necessary – all AI-assisted content before publication or distribution. Designate staff members responsible for fact-checking and ensuring accuracy of AI-assisted work. Policies Consider developing an AI policy that is transparently shared on your website. It can include commitments to mark all AI-assisted content with a label or footnote. Staff Training and Education Provide comprehensive training on AI capabilities, limitations, and risks. Ensure staff understand both the potential and the pitfalls of these tools. System Settings and Privacy Configurations Most platforms offer various privacy and data usage settings that users should review carefully. Common settings include options to "improve the model for everyone" (which may allow your conversations to be used in training data), memory features that save information across sessions, and data retention policies. For maximum privacy protection, users should generally disable data sharing for model improvement and turn off persistent memory features. However, even with privacy settings enabled, users should still avoid uploading confidential data or sensitive organizational information. Staying Current The AI landscape evolves rapidly, making it essential for organizations to stay informed about new capabilities and emerging risks. Reliable sources for ongoing AI security and capability updates include the National Institute of Standards and Technology (NIST) , though their resources may be affected by recent budget cuts. Academic institutions such as Stanford's Human-Centered AI Institute and MIT's Computer Science and Artificial Intelligence Laboratory publish research on AI safety and capabilities. Cost The adage “you get what you pay for” is particularly true with AI. The free versions of ChatGPT, Claude, Gemini and others all come with significant usage limitations, thanks to resource-intensive nature of the systems. All of these platforms offer paid tiers at around $20/month, which grant more time and extra features like folders/projects and, in some cases, the ability to opt out of having your chats used to train AI models.   While this monthly fee will feel steep to many, if you come to rely on AI tools for your daily work it is likely worth the cost. Privacy-Focused Alternatives Several platforms offer access to popular AI models through privacy-focused interfaces. Services like Duck.ai and Kagi's AI features provide access to leading language models while implementing stronger privacy safeguards, such as not storing conversation histories or using interactions for model training. However, users should still exercise the same caution regarding sensitive data, as queries are still be processed by the underlying AI models (such as ChatGPT, Claude, Llama, and Google’s Gemini) even if the intermediary platform doesn't retain the information. Understanding and Using AI Responsibly A guide for mission-driven organizations on how large language models work, where they help and fall short, and how to choose the right level of investment in AI tools. 1. What Is a Large Language Model, and Why Does It Matter? A large language model, or LLM, is the technology behind tools like ChatGPT, Claude, Gemini, and Copilot. It is a type of artificial intelligence trained to predict and generate human-like text by studying enormous amounts of existing writing. Think of an LLM as a very well-read assistant who has skimmed a huge share of everything ever published online, but who was never trained to fact-check what it read, and who cannot tell you where a specific idea came from. It generates responses by recognizing patterns in language, not by understanding or reasoning about the world the way a person does. How an LLM Is Built LLMs are trained on massive collections of text scraped and licensed from the internet and other sources, including: Websites, forums, and blogs Digitized books and published literature News articles and journalism Encyclopedic and reference content Other publicly available or licensed digital text The sheer scale of this training data is what makes an LLM feel capable and fluent. But that same data has a shape and a set of gaps, and those gaps show up in what the AI produces. Why This Creates Bias, Especially a Language and Cultural Bias The internet is not an even, representative sample of human knowledge and experience. A disproportionate share of the highest-quality, most abundant digital text is written in English and reflects perspectives, institutions, and cultural norms from wealthier, English-speaking, internet-connected parts of the world. As a result, AI tools tend to: Perform best in English and struggle more with other languages, dialects, and regional phrasing Reflect assumptions, norms, and framing common in U.S. and Western media and institutions Underrepresent perspectives, histories, and expertise from communities with less digital presence, including many communities that civil society and advocacy organizations serve Reproduce stereotypes or skewed framing present in the training data, sometimes subtly Groundwire's Advice: Because your organization's mission likely centers communities that are underrepresented in mainstream digital content, treat AI output about those communities with extra scrutiny. If a response feels generic, flattened, or slightly "off" about the population you serve, that is often the language and cultural bias in the training data showing through, not a neutral or authoritative answer. 2. Automation vs. AI: What's the Difference? These terms are often used interchangeably, but they describe different kinds of technology, and knowing the difference helps you pick the right tool for a task. Automation Automation follows fixed, pre-programmed rules to complete repetitive tasks the same way every time. It does not "think," generate new content, or adapt to unfamiliar situations. Automation is predictable and consistent by design. Examples: a scheduled email reminder, a form that auto-fills a donor's name into a thank-you letter, a spreadsheet formula, a chatbot that only answers from a fixed script of FAQs Strength: reliable, repeatable, easy to audit — it does the same thing every time Limit: cannot handle anything outside the rules it was given Artificial Intelligence AI, and generative AI in particular, produces new content or judgments by recognizing patterns rather than following fixed rules. It can handle open-ended requests, adapt its response to context, and generate something that did not exist before — but its output can vary and is not guaranteed to be accurate. Examples: drafting a letter to a policymaker, summarizing a 40-page report, brainstorming talking points, answering an open-ended question about your uploaded documents Strength: flexible, handles novel or complex requests, saves drafting time Limit: can be inconsistent, confidently wrong, and requires human review every time Groundwire's Advice: If a task is repetitive and the rules never change, look for an automation, not an AI tool — it will be more reliable and often cheaper. Reach for AI when the task is open-ended, requires drafting or synthesis, and you have time to review the output before it goes out the door. 3. Where AI Can Help — and Where It Has Limits How AI Can Help AI can help you save time by: Drafting content Rewriting text Summarizing information Organizing ideas Identifying patterns in large amounts of information Answering questions based on information it has access to Where AI Has Limits AI can be helpful, but it still needs human review because it does not: Understand the full situation or context Know whether information is accurate or complete Recognize what is fair, appropriate, or sensitive Replace your professional expertise, knowledge, or judgment 4. Why AI Responses Always Need Human Review When AI Sounds Confident AI can write responses that sound polished and confident, even when the information is incomplete, outdated, or wrong. That means you may see responses that: Sound accurate but include incorrect information Leave out important context or details Repeat or reinforce an assumption instead of correcting it Look professional even when they still need review When AI Creates False Information AI may create information that sounds real but cannot be verified or is not accurate. This is sometimes called a hallucination. This can happen when AI fills in missing details instead of saying it does not know. Be especially careful with: Numbers, statistics, or dates Citations, sources, or links Technical specifications or product details Policy, legal, or compliance information Specialized topics that require expert knowledge A Simple Way to Think About AI Think of AI as a helpful assistant that has seen many examples and can create a polished response quickly. However, AI may fill in missing details, make assumptions, or present information with more confidence than it should. Use AI to support your work, but do not rely on it as the final answer without reviewing the response first. Before You Use an AI Response Before copying, sharing, or acting on AI-generated content, ask yourself: Does this match what I know or what trusted sources say? Is any important context missing? Does the response sound more certain than the information supports? Have I verified key facts before using it? Remember: AI output should always be treated as a first draft, not a final answer. 5. How to Prompt AI Effectively The quality of an AI response depends heavily on the quality of the prompt. A simple framework — Context, Role, Instructions, and Tone — will get you a far more useful first draft. Context — share helpful background "I am reviewing a vendor proposal for a clinical operations team." Role — tell AI how to approach the task "Act as a neutral assistant helping summarize information." Instructions — say exactly what you want "Summarize the key offerings in three bullet points." Tone — describe how it should sound "Use clear, plain language suitable for a non-technical audience." 6. Should Your Organization Get an Enterprise AI License? Once staff start relying on AI regularly, organizations face a real decision: stay on free or individual paid accounts, or invest in an enterprise-tier license. For a large corporation this is a simple budget line. For a small civil society organization with a handful of staff and a tight budget, it is a genuine trade-off worth thinking through deliberately. The Case for an Enterprise License An enterprise or business-tier plan (as opposed to a free account or an individual $20/month subscription) typically adds: Data protection by default: inputs are excluded from model training and are not used to improve the underlying AI, which matters if staff are pasting in draft grant language, HR content, or program data Centralized admin controls: single sign-on, the ability to see (in aggregate) how the tool is being used, and the ability to immediately revoke access when someone leaves the organization Consistency and oversight: one sanctioned tool is far easier to govern than a dozen staff quietly using a dozen different free tools with unclear settings, sometimes called "AI sprawl" Higher usage limits and stronger models: fewer interruptions from rate limits, and often access to more capable versions of the AI Compliance support: audit logs, data retention controls, and documentation that can matter for funder requirements or legal review Why Many Small Organizations Hesitate Cost adds up fast: enterprise pricing is generally quoted per seat, often in a similar range to the ~$20/month individual tier but with minimum seat counts or contract terms — a real cost for an organization with five or ten staff and a lean budget Procurement overhead: contracts, data processing agreements, and admin setup take staff time that a small team may not have Uneven staff usage: if only two or three people use AI daily, paying for enterprise seats for everyone may not be worth it yet Free and individual tiers have improved: many platforms now let individual paid users opt out of having their chats used for training, closing some of the privacy gap with enterprise tiers, though admin oversight and audit tools still require the enterprise tier The Privacy and Security Angle For organizations serving vulnerable populations — immigrants, survivors, litigants, patients, or other communities where a data exposure could cause real harm — privacy and security should weigh more heavily than cost alone. On any tier, free or enterprise, staff should never enter names, addresses, case details, or other identifying or confidential information into an AI tool. Treat every AI input as if it could become public. Enterprise and business tiers generally guarantee, contractually, that inputs are not used for model training — free tiers may not offer this by default and require staff to manually adjust settings, which is easy to forget Centralized admin control means leadership can see which tool the organization is using and turn off access in one place, rather than discovering after the fact that AI sprawl left data scattered across accounts no one is tracking If your organization has real subpoena risk or works with especially sensitive information, a privacy-first tool built around end-to-end encryption is worth exploring alongside — or instead of — a mainstream enterprise plan (see Confer, below) Groundwire's Advice: Small organizations do not need to choose the most expensive option. A reasonable starting point is: pick one AI tool as your organization's standard, put it on individual paid accounts (with data-sharing settings turned off) while usage is light, and move to an enterprise or business tier once several staff are using AI daily or once you're regularly handling anything sensitive. Whatever tier you choose, pair it with a written AI policy so staff know what they can and cannot enter into the tool. Comparing Enterprise AI Vendors The table below compares the leading enterprise AI platforms most organizations are choosing between today. Feature ChatGPT Enterprise (OpenAI) Claude for Business (Anthropic) Gemini for Business (Google) Microsoft 365 Copilot Primary Strength Creativity, R&D, coding, content ideation Safe, reliable AI for complex reasoning & code Real-time web search & Google Workspace integration Deep integration with Microsoft 365 apps (Word, Teams, Outlook, etc.) Data Privacy Never trains on customer data; data is not used to improve models Customer data is not used for training; explicit privacy-first design Workspace data used contextually but not for model training; customers can disable AI data usage Customer data not used to train models; processed within the Microsoft cloud Admin Controls SSO, SCIM, role-based access, usage analytics Self-serve seat management, spend caps, usage & code analytics, compliance API Admin console, contextual grounding in Drive/Gmail, data regions, DLP controls Full M365 admin center integration, conditional access, sensitivity labels Security Encrypted in transit and at rest; data residency available; no OpenAI staff access to customer data Encrypted in transit and at rest; no training on inputs; audit logs and compliance API Encrypted in transit and at rest; DLP, context-aware access, data region controls Encrypted in transit and at rest; Microsoft Zero Trust model; integrates with Defender and Purview Workspace Integration SharePoint, OneDrive, Google Drive, Box, GitHub; custom app integrations via GPTs Google Workspace, GitHub, Slack via API or webhooks Native in Gmail, Docs, Drive, Meet, Calendar ("Gemini in Workspace") Fully embedded in Word, Excel, PowerPoint, Outlook, Teams, Loop Collaboration Features Shared GPTs, team knowledge base, chat with files (PDFs, spreadsheets) Projects & Artifacts (shared workspaces), team prompts & templates Real-time co-editing with AI suggestions, AI-generated meeting summaries in Meet Meeting recaps in Teams, smart scheduling in Outlook, collaborative drafting Compliance & Governance SOC 2, ISO 27001, GDPR; data residency options SOC 2, HIPAA, GDPR; compliance API for audit and retention FedRAMP, HIPAA, GDPR; data regions, eDiscovery Full M365 compliance suite; Purview, eDiscovery, sensitivity labels A Privacy-First Alternative to Watch: Confer A newer category of AI platform is emerging that is built around privacy as the foundation rather than an add-on feature. Confer, launched by Moxie Marlinspike (the creator of Signal), is the leading example. It's worth understanding, though it is not yet a like-for-like substitute for the enterprise platforms above. Confer is designed as a private, end-to-end encrypted AI assistant — the goal is that no one, not even Confer itself, can see what you say to it If Confer were subpoenaed, the government could not see what was shared with it — only that an account exists It runs on a small subscription model (roughly $35/month) rather than an enterprise per-seat contract, and currently has no team plan, admin console, API, or compliance documentation — meaning it is best suited to an individual staff member handling especially sensitive conversations, not yet a replacement for an organization-wide platform It does not currently support file or document uploads, or collaboration features, though this may change as the platform matures Groundwire's Advice: Confer is a great option for an individual staff member — an executive director, an attorney, or anyone regularly discussing sensitive matters — who wants a more private alternative to mainstream chatbots for one-off, non-collaborative use. It is not yet ready to serve as your organization's primary, organization-wide AI platform. We recommend watching how it develops and pairing it with, rather than substituting it for, your main enterprise tool. A Final Word AI is a genuinely useful tool for a resource-constrained organization — but it is a tool, not a replacement for your judgment, your relationships, or your mission expertise. Whatever combination of tools and license tiers you choose, the constant should be human review, a clear written policy, and staff who understand both what AI is good at and where it falls short. What is an Enterprise AI License? Does My Organization Need One? A companion guide to Understanding and Using AI Responsibly, for organizations weighing whether to move from free or individual AI accounts to an enterprise-tier license. Once staff start relying on AI regularly, organizations face a real decision: stay on free or individual paid accounts, or invest in an enterprise-tier license. For a large corporation this is a simple budget line. For a small civil society organization with a handful of staff and a tight budget, it is a genuine trade-off worth thinking through deliberately. The Case for an Enterprise License An enterprise or business-tier plan (as opposed to a free account or an individual $20/month subscription) typically adds: Data protection by default: inputs are excluded from model training and are not used to improve the underlying AI, which matters if staff are pasting in draft grant language, HR content, or program data Centralized admin controls: single sign-on, the ability to see (in aggregate) how the tool is being used, and the ability to immediately revoke access when someone leaves the organization Consistency and oversight: one sanctioned tool is far easier to govern than a dozen staff quietly using a dozen different free tools with unclear settings, sometimes called "AI sprawl" Higher usage limits and stronger models: fewer interruptions from rate limits, and often access to more capable versions of the AI Compliance support: audit logs, data retention controls, and documentation that can matter for funder requirements or legal review Why Many Small Organizations Hesitate Cost adds up fast: enterprise pricing is generally quoted per seat, often in a similar range to the ~$20/month individual tier but with minimum seat counts or contract terms — a real cost for an organization with five or ten staff and a lean budget Procurement overhead: contracts, data processing agreements, and admin setup take staff time that a small team may not have Uneven staff usage: if only two or three people use AI daily, paying for enterprise seats for everyone may not be worth it yet Free and individual tiers have improved: many platforms now let individual paid users opt out of having their chats used for training, closing some of the privacy gap with enterprise tiers, though admin oversight and audit tools still require the enterprise tier The Privacy and Security Angle For organizations serving vulnerable populations — immigrants, survivors, litigants, patients, or other communities where a data exposure could cause real harm — privacy and security should weigh more heavily than cost alone. On any tier, free or enterprise, staff should never enter names, addresses, case details, or other identifying or confidential information into an AI tool. Treat every AI input as if it could become public. Enterprise and business tiers generally guarantee, contractually, that inputs are not used for model training — free tiers may not offer this by default and require staff to manually adjust settings, which is easy to forget Centralized admin control means leadership can see which tool the organization is using and turn off access in one place, rather than discovering after the fact that AI sprawl left data scattered across accounts no one is tracking If your organization has real subpoena risk or works with especially sensitive information, a privacy-first tool built around end-to-end encryption is worth exploring alongside — or instead of — a mainstream enterprise plan (see Confer, below) Groundwire's Advice: Small organizations do not need to choose the most expensive option. A reasonable starting point is: pick one AI tool as your organization's standard, put it on individual paid accounts (with data-sharing settings turned off) while usage is light, and move to an enterprise or business tier once several staff are using AI daily or once you're regularly handling anything sensitive. Whatever tier you choose, pair it with a written AI policy so staff know what they can and cannot enter into the tool. Comparing Enterprise AI Vendors Each platform below is compared across the same seven attributes: primary strength, data privacy, admin controls, security, workspace integration, collaboration features, and compliance & governance. ChatGPT Enterprise (OpenAI) Primary Strength: Creativity, R&D, coding, content ideation Data Privacy: Never trains on customer data; data is not used to improve models Admin Controls: SSO, SCIM, role-based access, usage analytics Security: Encrypted in transit and at rest; data residency available; no OpenAI staff access to customer data Workspace Integration: SharePoint, OneDrive, Google Drive, Box, GitHub; custom app integrations via GPTs Collaboration Features: Shared GPTs, team knowledge base, chat with files (PDFs, spreadsheets) Compliance & Governance: SOC 2, ISO 27001, GDPR; data residency options Claude for Business (Anthropic) Primary Strength: Safe, reliable AI for complex reasoning & code Data Privacy: Customer data is not used for training; explicit privacy-first design Admin Controls: Self-serve seat management, spend caps, usage & code analytics, compliance API Security: Encrypted in transit and at rest; no training on inputs; audit logs and compliance API Workspace Integration: Google Workspace, GitHub, Slack via API or webhooks Collaboration Features: Projects & Artifacts (shared workspaces), team prompts & templates Compliance & Governance: SOC 2, HIPAA, GDPR; compliance API for audit and retention Gemini for Business (Google) Primary Strength: Real-time web search & Google Workspace integration Data Privacy: Workspace data used contextually but not for model training; customers can disable AI data usage Admin Controls: Admin console, contextual grounding in Drive/Gmail, data regions, DLP controls Security: Encrypted in transit and at rest; DLP, context-aware access, data region controls Workspace Integration: Native in Gmail, Docs, Drive, Meet, Calendar ("Gemini in Workspace") Collaboration Features: Real-time co-editing with AI suggestions, AI-generated meeting summaries in Meet Compliance & Governance: FedRAMP, HIPAA, GDPR; data regions, eDiscovery Microsoft 365 Copilot Primary Strength: Deep integration with Microsoft 365 apps (Word, Teams, Outlook, etc.) Data Privacy: Customer data not used to train models; processed within the Microsoft cloud Admin Controls: Full M365 admin center integration, conditional access, sensitivity labels Security: Encrypted in transit and at rest; Microsoft Zero Trust model; integrates with Defender and Purview Workspace Integration: Fully embedded in Word, Excel, PowerPoint, Outlook, Teams, Loop Collaboration Features: Meeting recaps in Teams, smart scheduling in Outlook, collaborative drafting Compliance & Governance: Full M365 compliance suite; Purview, eDiscovery, sensitivity labels A Privacy-First Alternative to Watch: Confer A newer category of AI platform is emerging that is built around privacy as the foundation rather than an add-on feature. Confer, launched by Moxie Marlinspike (the creator of Signal), is the leading example. It's worth understanding, though it is not yet a like-for-like substitute for the enterprise platforms above. Primary Strength: Private, end-to-end encrypted AI assistant — the goal is that no one, not even Confer itself, can see what you say to it Data Privacy: If Confer were subpoenaed, the government could not see what was shared with it — only that an account exists Admin Controls: None currently — no team plan or admin console Security: End-to-end encryption is the core design principle, not an add-on Workspace Integration: None currently — no file or document upload support Collaboration Features: None currently — built for individual, non-collaborative use Compliance & Governance: No compliance documentation available yet Cost: Roughly $35/month, individual subscription (not a per-seat enterprise contract) Groundwire's Advice: Confer is a great option for an individual staff member — an executive director, an attorney, or anyone regularly discussing sensitive matters — who wants a more private alternative to mainstream chatbots for one-off, non-collaborative use. It is not yet ready to serve as your organization's primary, organization-wide AI platform. We recommend watching how it develops and pairing it with, rather than substituting it for, your main enterprise tool. A Final Word Whatever combination of tools and license tiers you choose, the constant should be human review , a clear written policy, and staff who understand both what AI is good at and where it falls short. It is also important to keep in mind, that even with an enterprise license, that AI company still stores your data, similarly to the way Google or Microsoft stores your data. There is no single right answer for every organization — the right tier depends on your budget, how much staff already rely on AI, and how sensitive the information you handle tends to be.