Public Offices, Agencies and Campaigns

Give your office or campaign an AI policy and a trained staff.

We help you write a clear AI policy, train your staff and volunteers to follow it, and plan your response to deepfakes before you need to.

Nonpartisan. We do not work for or against any party or candidate.

Who It’s For

Two kinds of public buyers

A government office and a campaign both work in public, but they face different rules and different risks. The policy is built for the one you are.

Government offices and agencies

Elected offices, agencies, departments, and legislative and county offices. You work under public rules, and your data belongs to the public you serve.

  • A policy that fits under your government’s rules

    Your government may already have an AI policy or IT rules. Your office policy sits under them. We read them first and align to them. We do not replace them.

  • Constituent data

    Case files, requests and contact records stay out of AI tools unless your rules and your approved tools allow them.

  • Prompts and outputs may be public records

    In some jurisdictions, what staff type into an AI tool and what it returns may be public records. Check with your counsel or records officer on how yours are kept and released.

  • Approved tools only

    Staff use the tools your IT and procurement rules approve, and they know which ones those are.

Campaigns and party organizations

Campaign committees, party organizations and caucuses. You move fast, you rely on volunteers, and a public mistake costs you in the news cycle.

  • Staff and volunteer turnover

    People join late and leave early. The rules have to be short, clear and easy to teach again to each new person.

  • Voter and donor data terms

    Voter files and donor lists often come with use terms from whoever supplied them. The policy keeps that data out of any tool those terms do not allow.

  • AI disclosure rules

    Rules on disclosing AI in political ads and messages vary by jurisdiction. Check with your counsel which ones apply to you.

  • Deepfakes of candidates and staff

    Fake audio, video or images of a candidate or a staff member. Decide who responds, and how, before one appears.

The Package

What we deliver

The policy comes first. The training is how your people learn to follow it.

1

Leadership briefing

For leadership

For the official, chief of staff, campaign manager or party director. We cover which AI tools your people already use, where the risks are, and which decisions the policy needs from leadership.

2

Your AI policy

Traffic-light method

A written AI policy for your office or campaign. It sorts your data into RED, YELLOW and GREEN, names the approved tools, and says who to ask. Written for review by your counsel before you adopt it.

3

Staff and volunteer training

Hands-on

Your people practice on the tools you approve, applying the traffic light to their own tasks: drafting replies, summarizing public documents, preparing briefing notes.

4

Deepfake response playbook

Rehearsed with staff

A written plan for what your team does when a fake image, video or recording of your people appears. Details below.

The Traffic-Light Method

Every piece of data gets a color. The color decides where it can go.

RED

RED data never goes into any AI tool.

Examples: Non-public constituent case details. Voter file exports or donor records outside approved systems.

YELLOW

YELLOW data goes only into an approved enterprise AI tool covered by a signed agreement (for example a data-processing agreement, or a BAA for health data).

Examples: Internal drafts not yet released, staff meeting notes, routine internal correspondence.

GREEN

GREEN data can be used in any approved tool.

Examples: Published press releases, public meeting agendas, reports already made public.

Your office or campaign decides which data falls in which tier. We help you sort it and write it down.

Deepfake Response Playbook

A plan for the day a fake of your people appears

A fake image, video or recording of a candidate, official or staff member can spread before anyone checks it. The playbook sets out who does what, in order, so your team does not have to invent a process under pressure.

  1. 1

    The response team and who decides

    Names who is on the team, who makes the call, and who speaks publicly. Includes backups for evenings and weekends.

  2. 2

    How staff report a suspected fake

    One clear way for any staff member or volunteer to report something suspicious, and who receives the report.

  3. 3

    How to preserve evidence

    Save the file, the URL, the time it was found, and screenshots, before the post is edited or removed.

  4. 4

    How to verify with the person depicted

    Check directly with the person shown or heard, and with the people who would know where they were, before you respond.

  5. 5

    Pre-approved holding statements

    Short statements approved in advance, so you can say something accurate quickly while the facts are checked.

  6. 6

    Who to contact, and when

    When to report to the platform, when to call your counsel, and when, if relevant, to contact law enforcement or election officials.

  7. 7

    Post-incident review

    After it is over: what worked, what did not, and what changes in the playbook.

We rehearse the playbook with your staff in a tabletop exercise: a walk-through of a realistic scenario, so people know their roles before a real incident.

Whether a platform removes content, and which legal options you have, depends on the platform and your jurisdiction. Your counsel advises on those.

Investment

Pricing and delivery

Starts at $7,500 (policy and rollout) or $12,500 (two-day on-site training for up to 25 people).

Travel terms are set in your written proposal. We confirm scope and price with you on the discovery call, based on your staff size and what you need.

Delivery

In person in the Pacific Northwest first. Travel and virtual delivery are available.

Timing

Campaigns often have the least time to act in the weeks close to an election. New offices and new staff face the same squeeze. A plan made before crunch time is easier to follow when it counts.

Questions

What public buyers ask first.

Are you affiliated with a political party?

No. The Institute is nonpartisan. We do not work for or against any party or candidate, and we teach every client the same method. We teach AI policy and safe AI use; we do not write campaign strategy or messaging.

Is this legal advice?

No. We help you write a clear AI policy and train your people to follow it. Laws on AI, public records, elections and data vary by jurisdiction. Have your counsel review the policy before you adopt it.

Can you work alongside our IT department or our state or county AI policy?

Yes. If your government already has an AI policy or IT rules, your office policy sits under them. We read them first and align to them; we do not replace them. Your IT staff and records officer are welcome in the work.

Do you train volunteers?

Yes. Campaign and party training covers staff and volunteers. Because volunteers change often, we keep the rules short so your team can brief new people with the same material.

How long does it take?

It depends on your size and on how quickly leadership can make decisions. On-site team training is two days; the other pieces are scheduled around your leadership. We set the schedule with you on the discovery call.

Put a plan in place before you need it.

The Institute is the neutral teacher: we teach AI policy and safe AI use, and we do not build AI systems.

Book a discovery call→