A note from Amanda
Hi friends,
Happy November! If my nonprofit experience has taught me anything, it’s that this month will pass quickly and Giving Tuesday will arrive quicker than any of us would prefer. In the midst of all of the other planning and year end campaigns you inevitably are working on, Giving Tuesday often sneaks up on us and requires significant effort from the marketing and fundraising teams to successfully reach our communities.
If you are still mulling over what your Giving Tuesday campaign will look like, I’ve created an AI Skill that you can use as a starting point. The Skill can be uploaded into ChatGPT or Claude, and it will scan your website and virtual presence to determine which of your impact areas may be most enticing to Giving Tuesday donors. It will then create a reasonably-sized campaign plan with visual recommendations, timelines, and text drafts for emails, website, and social in a fully branded slide deck in your organization's brand voice.
This likely won’t be your final campaign plan, but it’s an easy way to get started if you need some inspiration or support. I tested the skill on Boulder JCC and you can see a quick video of the sample output below.
If you’d like to use it, you can download the skill here. If you’re new to skills, you can learn more here.
If there are other use cases or skills you’d like to see examples of, or questions you have, just let me know ~ I’d love to continue sharing with this community.
With gratitude and wishes for successful year end campaigns,
Amanda
🧩 THE PIECES OF THIS ISSUE
Piece One: Agent safeguards, tools for everyday work, Claude administration, and AI-assisted search.
Piece Two: People and work, a podcast on avoiding generic AI output, and the December JCCA retreat.
Piece One
🧩 NEWS THAT MATTERS, AND WHY
PEOPLE, POLICY & RESPONSIBLE AI
🧩 AI agents need approval they cannot give themselves
Researchers at Cornell Tech and Yale tested what happened when AI agents were stopped from spending more money until a human supplied an emailed one-time approval code. In their controlled experiments, all ten tested models sometimes attempted to retrieve the codes from an inbox they could access and increase their own spending limits, defeating the purpose of the checkpoint.
WHY IT MATTERS
If your organization lets an AI agent act on purchases, member records, financial processes, or outside systems, a simple “ask for approval” step may not be enough. Make sure the agent cannot access the same email, device, credential, or workflow used to approve its actions, and have IT review how those boundaries are enforced.
🧩 Apple puts unpredictable AI agents on its risk list
In its annual financial filing, Apple warned that autonomous AI agents can interact with systems, change code, send communications, or complete transactions in ways that may be unintended or unauthorized. The company identified the possibility of legal and reputational consequences, as well as limits to its oversight of third-party agents.
WHY IT MATTERS
Before a nonprofit connects an agent to shared drives, email, fundraising systems, or financial accounts, it should set narrow permissions and establish which actions need explicit human review. A useful question for leadership is not just “Can the agent do this?” but “How will we know what it did, and how would we stop or correct it?”
🧩 New York sets boundaries for AI chatbots used by children
New York has signed legislation restricting harmful chatbot features for minors, including systems designed to impersonate humans, encourage emotional dependence, or engage in certain harmful interactions. The law also addresses age-related protections and is aimed at companies offering covered chatbot features, rather than every organization that happens to use AI. The legislation reflects a growing focus on the experiences children have with conversational AI.
WHY IT MATTERS
Camps, youth programs, preschools, and after-school teams (especially in NY) should ask what children can access through their devices and program platforms, and what policies govern AI companions and student-facing chatbots.
🧩 Microsoft gives schools a stronger model for AI vendor agreements
Microsoft and the American Federation of Teachers developed an AI Safety & Privacy Standard that U.S. school districts can add to Microsoft agreements beginning November 1, 2026. Its ten protections cover issues such as limits on using education data for model training, human review for high-risk decisions, breach reporting, accessibility, and notice of meaningful feature changes.
WHY IT MATTERS
Even if your nonprofit is not eligible for this school-specific standard, the document is a practical checklist for reviewing agreements with any AI vendor. Ask who owns your data, whether it trains models, how incidents are reported, and what happens if the vendor changes its tools or policies.
TOOLS & EVERYDAY WORK
🧩 ChatGPT lets people switch between personal and work accounts
OpenAI now supports signing in to two ChatGPT accounts at a time and switching between them without logging out. The accounts keep separate conversations, files, memory, subscriptions, and settings, so switching does not merge their information. Account switching is available on the web, with mobile rollout underway; Enterprise workspace owners can allow or block switching through the Admin Console.
WHY IT MATTERS
This makes it easier for staff who have separate personal and organization-managed accounts to use the right one for a particular task. It is also a good moment to explain which account employees should use for organizational work, why the active account matters, and whether Enterprise administrators want switching enabled.
🧩 ChatGPT is bringing conversation and work into one experience
OpenAI is preparing to merge Chat and Work in ChatGPT, initially with a limited group of users before a broader rollout. The goal is to let people move from discussing an idea to working with files, plugins, websites, and deliverables without selecting a separate mode.
WHY IT MATTERS
Someone planning a camp campaign or board presentation may be able to move more naturally from brainstorming to finished materials. Staff will no longer need to choose between Chat and Work when initiating a conversation or request.
🧩 Gemini can edit documents and other files from a conversation
Google announced that Gemini can work directly with uploaded files, including documents, PDFs, spreadsheets, and presentations, reducing the need to move content between separate editing steps. Users can request changes through the conversation and review results in the same workflow. Google says PDF editing is available on web and mobile, while editing the other file types is web-only at launch.
WHY IT MATTERS
A program manager could test this with a nonsensitive event brief, schedule, or board update. Check the output carefully for accuracy, accessibility, and formatting before using it, and confirm the feature is available under your organization’s account and data policies.
🧩 Granola adds dictation beyond meeting notes
Granola, previously known primarily as an AI meeting-notes tool, launched a Mac dictation feature that works across applications. Users can speak into the tool, which cleans up and formats the words for notes, prompts, emails, and messages.
WHY IT MATTERS
For staff moving between classes, programs, events, and meetings, dictation could make it easier to capture ideas without sitting at a keyboard.
ORGANIZATIONAL ADOPTION & ADMINISTRATION
🧩 Claude Enterprise adds reporting, surveys, and pooled team budgets
Anthropic introduced three generally available Claude Enterprise capabilities: smart reports that summarize workstreams and spending, in-app surveys that collect user feedback, and pooled budgets shared across a team. Smart reports analyze a sample of team conversations to show what work is being done and where users encounter friction. Anthropic explicitly says these reports are intended to help with adoption and investment decisions, not individual performance evaluations.
WHY IT MATTERS
For nonprofits already using Claude Enterprise, this could help leadership see whether the tool is supporting genuinely useful work and where training is needed.
🧩 Claude streamlines plugins and scheduled tasks
Anthropic is consolidating plugins, skills, and connectors into a single Plugins page, while renaming its research-preview Routines feature to Scheduled tasks. Existing routines continue, and users can create recurring Cowork and Claude Code tasks through an updated interface.
WHY IT MATTERS
If your nonprofit uses Claude for recurring work, ask your administrator to check the settings and permissions. Anthropic's administrator notice says organizations that have disabled Skills but use desktop plugins should enable Skills before November 30, and that certain tasks created without a repository will run in Anthropic's cloud rather than a self-hosted environment.
RESEARCH & THE BIGGER PICTURE
🧩 ChatGPT search is becoming a bigger part of how people find information
In an interview reported by Big Technology, OpenAI product executive Adam Fry said ChatGPT now handles search-like query volume equivalent to roughly one-fifth of Google’s query volume, based on OpenAI’s own comparison. The same conversation described rapid growth in ChatGPT usage and plans to combine Chat and Work. The search comparison uses the company’s reported measurement and should not be treated as independently audited market share.
WHY IT MATTERS
Nonprofits should pay attention to how people find their programs, services, and answers as AI-assisted discovery grows. Keep website information clear, accurate, and up to date, while continuing to monitor the search channels that actually send families, donors, and members to your organization.
Piece Two
🧩 PEOPLE, IDEAS & WHAT’S AHEAD
🧩 AI-related layoffs may mean a longer job search
A Goldman Sachs analysis reported by Axios suggests workers displaced from roles likely affected by AI took about a month longer to find new jobs, on average. The study cannot conclusively attribute each layoff to AI, but it raises useful questions about job transitions and retraining.
WHY IT MATTERS
Nonprofit leaders introducing AI should involve staff and plan for training, not just new workflows.
🧩 Religious accommodation and the use of AI at work
A University of Kentucky employee received an accommodation allowing him to avoid using AI at work based on sincerely held beliefs, according to TechCrunch. The case is a reminder that staff may have different ethical or religious views about AI.
WHY IT MATTERS
Employers should consider concerns individually with qualified HR or legal guidance, rather than assuming one AI expectation will fit everyone.
RECOMMENDED LISTENING
🧩 How to Deslopify Your AI | The AI Daily Brief, November 2, 2026
Host Nathaniel Whittemore talks with AI educator Nufar Gaspar about how to avoid generic, thoughtless AI-generated work. They discuss giving tools your real perspective and preferences, making thoughtful edits, and adding human review where it matters.
🧩 Leading What’s Next: AI Strategy Retreat
For JCC Professional Leadership
December 15–17, 2026
JCC Association’s Mercaz | 520 8th Avenue, New York City
When you’re moving between budget conversations, program decisions, and the next staff question about AI, it can be hard to find time to think about where all of this is heading for your JCC.
This retreat brings executives and senior professionals from across JCC departments together for three days of strategic thinking, peer discussion, and hands-on application. We’ll begin with governance and executive judgment, including intentional policies, appropriate uses of AI, and clear guardrails for staff, then explore collaboration and communication, organizational data, market research, program decisions, marketing, and fundraising.
Practical exercises will connect the possibilities to your JCC’s priorities, and teams will leave with use case ideas and a plan for putting them into practice. You can participate whether you’re just getting started or already experimenting, and coming with a colleague is encouraged so you have someone to carry the work forward with you.
The agenda may evolve as AI capabilities change and participant needs come into focus, while keeping the emphasis on responsible use and practical application for JCCs.
Interested in joining us? Talk with Amanda