The $100 Million Tell
Why the AI Safety Fight Is Happening in the States
Two announcements caught my eye this week and they should catch yours too. Meta is launching a California super PAC to elevate light‑touch AI candidates statewide, with “tens of millions” committed, and a separate “Leading the Future” super PAC network has at least 100 million dollars behind it from a16z, OpenAI’s Greg Brockman, and others. These vehicles are explicitly designed to tilt California and a small set of other big states toward industry‑preferred rules (or lack thereof).
This isn't a sign that reasonable state AI safety measures are doomed. It’s precisely the opposite.
You don’t drop $100 million to lobby if the debate is already settled in your favor. The Senate’s 99–1 vote in July to kill a ten‑year federal moratorium on state AI laws already told us the same thing: The real fight over AI safety is happening at the state level. And those fights are very much winnable.
But if we’re going to win them, we need to start now, and we don't need nine-figure budgets to do it.
The Opportunity We're Missing
Funding and organization for on-the-ground state networks have lagged significantly behind other AI safety priorities. Safety aligned organizations have heavily invested in training fellowships, alignment research, and white-paper production, and drastically under-invested in the unglamorous craft that actually changes the law.
This is a strategic error that needs immediate correction.
Longtermist funders should see the opening immediately. Work on red-teaming, formal methods, and international governance theory matters, but the pipes that carry assurance and accountability are clogged. You cannot design governance for a black box you do not have the right to open, or may not even know exists.
The low-hanging fruit that states can pass this session — transparency obligations, protected whistleblower channels, and third party evaluations —forces information into the open. This generates the disclosures, incident data, and institutional relationships that long-run governance requires. The investment in the present secures the data needed for the future.
For short-timeline believers who worry transformative AI could arrive this decade, waiting for a sweeping federal omnibus bill or hoping for an improbable global pause borders on negligence. State-level wins might seem incremental, even trivial, given existential stakes, but they’re our most tangible hedge against looming risks. They create levers of accountability, data points of exposure, and footholds of trust. The alternative is sitting idle, or worse, becoming a caricatured foil used by anti-regulation forces.
Yes, building coalitions is messy. Legislative schedules offer no mercy. Theory alone doesn’t move committee chairs. But these unglamorous victories matter enormously. They establish a baseline from which stronger protections can grow.
The Strategy: Leverage, Pragmatism, and Speed
To counter organized, heavily funded opposition, we cannot match them dollar for dollar. We must outmaneuver them using political leverage, pragmatic coalition building, and modern communication strategies.
First, lean into the California Effect. You don’t need uniform laws in every state. Winning decisive battles in influential markets like California sets national defaults. Industry lobbyists know this. That’s exactly why the money announced this week is laser-focused on California, New York, Illinois and Ohio. Our smartest play mirrors theirs: target these states, set benchmarks, and let national practices adapt accordingly. Be nimble enough to move personnel between states as needed. The current California and New York bills may not be perfect, but as I wrote about before, they are a great step in the right direction.
Second, resist the purity trap. "AI safety" can't become a partisan or ideological jersey. The Senate’s vote against a decade of state preemption was a cross‑ideological moment that kept the field open. You want both Steve Bannon’s War Room and Pod Save America to amplify your objectives. Currently, reasonable AI safety regulations are supported by majorities in both parties. The goal is to grow this base, not shrink it. Save the sharp theoretical debates for times and arenas where you can win them. The goal now is passage.
Third, embrace agile media strategies. Do not confuse large checkbooks with strategic inevitability. The "old playbook" of massive local TV buys is failing and the media market is tilting toward agility. Local TV is soft, connected TV and targeted digital are growing, and earned media still compounds if you are interesting and people care about the message. Use free and earned media where you can and budget for what people actually watch.
If you need a political case study, look at New York in 2025. Zohran Mamdani built a front‑running citywide campaign on an earned‑first, creator‑driven strategy that prioritized short video, podcasts, and relentless newsmaking over the old broadcast playbook. Even critics framed their attacks around his social media reach, which tells you where the attention lived. Our audience may be different. The attention dynamics are not.
A Blueprint for Effective Advocacy
Winning in the states does not require a massive organizational footprint. It requires a lean, specialized team designed for speed, local expertise, and operational discipline. AI governance may be new but the state advocacy playbook is not. Here is how you structure it.
Run a hub‑and‑spoke model. Centralize everything that benefits from repetition. Creative assets, web properties, CRM, analytics dashboards, legal templates, and training live in the hub and ship as versioned kits. Localize what depends on trust and tacit knowledge. Media relationships, coalition care, committee etiquette, and in‑building choreography belong to your in‑state communicator and your contract lobbyist.
Keep your core state team small and mobile: a State Director, a Policy Analyst, and (possibly) an organizer who can do coalitions and field. Add one digitally native, in‑state Communications Lead with real press relationships. Utilize grass roots passion to build up a volunteer-base. Buy deep building access through a respected bi-partisan lobbying firm (or two), not headcount. This can be achieved with surprisingly modest budgets and the hybrid gives you speed and flexibility without sacrificing local nuance.
Now Is the Moment
The influx of industry money confirms the stakes. The goal now is not to pass everything everywhere. It is to pass enough of the right things in the right places that practice changes everywhere. This strategy allows us to shape the national baseline faster and more efficiently than waiting for federal action.
The industry’s big spenders have loudly announced their arrival. It’s time we do the same.



Great article. Wondering if you could point me in the direction of good places for people to volunteer?