A companion to Paper III for advocacy organizations. The C3, C4 and PAC distinction has no candidate equivalent, and neither does a giving circle — this is the material that applies only to the organizations that publish the record.
Most progressive organizations operate across multiple legal entities — a 501(c)(3) for educational and research activities, a 501(c)(4) for issue advocacy, and a PAC for direct electoral activity. Each entity has different legal constraints on what it can say, who it can advocate for, and how it can spend resources. The AI era adds a dimension to this structure that few legal counsels have yet addressed: the digital footprint of each entity contributes to the AI information environment in ways that were not anticipated when the legal frameworks were designed.
The 501(c)(3) is the cleanest entity for AI information environment work. Educational and research content — policy analysis, legislative history, issue explainers, data publications — is exactly the kind of content AI systems tend to treat as most authoritative. A well-structured C3 research library, consistently maintained and schema-marked, can become a dominant source in AI outputs on its issue areas without raising any electoral advocacy concerns. This is education in its most literal sense: making accurate information available and accessible.
The 501(c)(4) issue advocacy entity can go further — explicit issue positions, legislative scorecards tied to votes on specific bills, organizational endorsements of policy positions — without engaging in direct electoral advocacy. The key is that positions are tied to issues and votes, not to electoral outcomes. An LCV scorecard that rates senators on their environmental votes is issue advocacy, not electoral advocacy, even if its practical effect is to help voters evaluate candidates. The AI information environment work is straightforward: make that scorecard structured, machine-readable, and consistently maintained.
The PAC can be most explicit — direct candidate endorsements, voter guides, electoral recommendations — but it must manage the separation between its activities and those of the C3 and C4. In the AI context, the practical concern is consistency: the PAC's candidate endorsements should be factually aligned with the C4's legislative scorecards and the C3's issue research. If they tell different factual stories, AI systems detect incoherence. If they tell the same factual story through different lenses appropriate to each entity's purpose, AI systems see a coherent, multi-source confirmation.
The legal framework that governs C3/C4/PAC operations was designed for a media environment where message control was the primary concern. In the AI environment, factual coherence across entities is the primary concern. A C3 research report, a C4 scorecard, and a PAC endorsement that all reference the same senator's votes on the same legislation, in factually consistent terms, create a powerful coherent signal for AI systems — while staying firmly within each entity's legal lane. The coordination that is prohibited is coordination of electoral activity. Coordination of factual accuracy is not prohibited. It is good practice.
This paper does not constitute legal advice, and organizational counsel should be involved in any AI optimization strategy that touches on C3/C4/PAC coordination questions. The important point for this conversation: the AI information environment dimension of entity structure is a genuinely new question that organizational counsel is unlikely to have worked through in advance. Briefing them on the specific question — how do our three entities' digital footprints interact in AI outputs, and what coordination of factual content is permissible — is a conversation worth having before you act, not after.
Giving circles — informal networks of aligned donors coordinating their political and philanthropic investments — occupy a distinctive position in the progressive ecosystem. They are not membership organizations. They do not publish scorecards or issue press releases. They operate with a light public footprint. And that light footprint, counterintuitively, gives them a specific advantage in the AI information environment.
Giving circles fund the research, the advocacy, the candidate development, and the infrastructure that produces the AI-visible content. They are, in a structural sense, upstream of the information ecosystem. The organizations they fund are the ones whose scorecards, endorsements, and research show up in AI outputs. Giving circles that understand this can use their investment decisions to deliberately strengthen the information ecosystem their supported candidates need to navigate.
Fund the infrastructure, not just the message. Organizations that build durable, structured, machine-readable archives of political information — legislative records, endorsement histories, issue research — are building assets that compound over time. A scorecard published in 2018 and still accessible in 2026 is contributing to the AI information environment eight years later. Funding organizations that maintain this kind of institutional infrastructure is funding something that has persistent, compounding value in the AI era.
Evaluate grantees on AI visibility, not just earned media. A traditional communications evaluation asks: how many press placements did the organization generate? How many people opened the email? A forward-looking evaluation asks: when voters ask AI about this organization's issue areas, is the organization's research reflected? Are the candidates it supports characterized accurately? These are measurable outcomes — they simply require new measurement practices.
Coordinate on factual record, not just messaging. Giving circles that fund multiple organizations working on adjacent issues have the ability to encourage coordination on the factual record across their portfolio. Not messaging coordination — each organization maintains its own voice and emphasis — but factual alignment: the same primary sources cited, the same legislative records referenced, the same data points used consistently. This is the kind of ecosystem coherence that AI systems tend to respond to with confident, accurate answers.
"The organizations that are winning in AI outputs are not necessarily the ones with the most resources. They are the ones with the most structured, coherent, consistently maintained information. Giving circles can fund exactly that."
The giving circle that takes AI information environment seriously as an investment lens is not changing what it funds — it is changing how it evaluates what it funds and what it asks of its grantees. That is a relatively low-friction shift that could have substantial impact on the effectiveness of the entire portfolio.