The people making AI make sense

Progressives for AI

Issue #10 · April 2026

Spotlight · News · Put AI to Work · Looking Ahead

In this issue

A TikTok creator is doing the AI translation work the industry refuses to do. Send her playlist to your volunteers.
California is one vote away from banning AI-only firing decisions. Your group chat should hear about this bill.
A small Vermont newsroom just negotiated AI protections other unions can copy tomorrow.
Put AI to work: point it at your Google Analytics and find out what your website is actually doing.

Creator spotlight: Sky Speirs

@skyspeirs on TikTok, "Solving the AI Power Problem"

Sky is an energy and climate educator who posts short videos about how AI actually gets powered, cooled, and built. Who pays for it. What the alternatives look like. Why "AI is bad for the planet" keeps missing the real fight. It's the kind of translation work the public conversation has been starved of.

A few of her hooks from the current playlist:

"Who should pay for AI energy costs: you or big tech?" Texas just passed a law requiring large energy users to pay minimum transmission charges. If the largest users covered the state's $33 billion grid investment, the average household would save $200 a year.
"Will AI break the grid? Short answer: No." The argument: AI doesn't have to be an energy hog. With flexibility, data centers can consume power when clean energy is abundant and ease off when it's not. "We have the headroom, we have the tech, we have the code, and we just need archaic rules to change."
"Data centers don't need to use a lot of water." Most facilities use evaporative cooling because it's cheap, not because it's good. Liquid cooling cuts water use by up to 90%. "Water-intensive data centers aren't a given. They're a design problem."

Sky's refrain across the series: "We created the problem so we can fix it." That's about as PFAI a sentence as exists on the internet.

Watch the playlist. If you do community education on AI, climate, or energy policy, her videos are a ready-made resource. Send them to your volunteers. Share them in your coalition Slack. Quote her at your next planning meeting.

AI News Roundup

A Vermont newsroom just wrote the AI union contract other locals should copy

What happened: On April 1, the VTDigger Guild ratified a four-year contract with the Vermont Journalism Trust that includes some of the most concrete AI protections yet negotiated in a small-newsroom unit. Among the provisions: management must give 60 days' notice before deploying any new AI system, the union has the right to negotiate over effects, laid-off workers affected by AI-driven restructuring get a minimum of 12 weeks of severance, and reporters can withhold their bylines from any story where AI tools played a substantive role. The contract also includes explicit language that "generative AI tools do not adequately substitute for human judgment" in journalism.

Why this matters: VTDigger is not a giant bargaining unit, and that's exactly why the contract matters. Small units often get held up as places where AI protections are impossible to negotiate. Here's a template that proves otherwise. Every clause is specific, enforceable, and portable. Notice periods, effects bargaining, severance floors, byline rights. Pick the ones that fit your shop and bring them to your next negotiation.

What you can do

If you're in a newsroom union, share this contract with your bargaining committee. If you work with unions outside of journalism, the principles translate: any workplace adopting AI tools can demand notice, effects bargaining, and protections against AI-driven layoffs.

Lawmakers are quietly using AI to draft bills, without telling anyone

What happened: Transformer News reported this week that state and federal legislators are increasingly using AI-powered drafting tools from companies like FiscalNote and Vulcan Technologies to write and analyze legislation. Virginia and Vermont have both started piloting these tools. The catch: there are virtually no disclosure requirements. Lawmakers can use AI to draft bills, summarize hearings, and analyze amendments without ever telling constituents, colleagues, or opponents.

Why this matters: This is a transparency fight with a clean progressive frame. The tools themselves aren't the problem. AI can make understaffed legislative offices faster and more effective, which is good for democracy. The problem is equity. If premium AI drafting tools cost $50,000+ per seat and corporate lobbyists use them to shape legislation while under-resourced advocacy groups can't, the playing field tilts further toward the people who already write the rules.

What you can do

Push your state legislature for AI-use disclosure standards that match what we already require for lobbying contacts. The ask is specific: any bill substantially drafted, edited, or analyzed using AI should carry a disclosure in the legislative history. That's not anti-technology. That's the same transparency we demand from every other tool of legislative influence.

In brief

OpenAI is buying a personal finance startup. OpenAI announced this week it's acquiring Hiro, a personal-finance AI startup, signaling that financial-planning features are headed to ChatGPT. Before that rolls out, someone needs to answer a basic question: what standard of care does an AI chatbot owe to a user asking where to put their retirement savings? Human financial advisors operate under fiduciary rules for a reason. If an AI tool is going to give financial advice to hundreds of millions of people, the advice should meet the same standard. Push your state financial regulator to start the conversation.

Progressive AI Win

California's "No Robo Bosses Act" is one vote away from becoming law.

Senate Bill 947, the California No Robo Bosses Act, cleared the Senate Labor Committee on April 8 and is headed for a full Senate vote on April 20. The bill would prohibit employers from relying solely on AI systems to fire, discipline, or demote workers. It requires meaningful human oversight of any automated employment decision, mandates that workers be given notice when AI is used in a personnel action, and creates a path to challenge decisions made primarily by algorithm.

The bill is backed by the California Labor Federation, SEIU California, and a coalition of worker advocacy groups. Tech industry lobbyists are pushing back, but the committee vote was a strong signal of momentum.

If you're in California: contact your state senator. The California Labor Federation has a simple action form. One email, one phone call, one minute. This is a winnable fight in front of us right now.

If you're not in California: this is the model bill. If SB 947 passes, it will be the first state-level ban on algorithmic firing in the country, and every other state will have a template. Start talking to your state legislator now about introducing a companion bill in your session.

Put AI to Work

Practical ways progressives can use AI this week

Ask AI what your website is actually doing

Most progressive organizations have Google Analytics installed and almost never look at it. The reports feel intimidating, the interface is dense, and nobody has an afternoon to spare. AI can fix that in about 20 minutes.

Step 1: Pull three reports from GA4. Log in to Google Analytics, go to Reports in the left sidebar. Export these three as CSV (the share icon in the top right of each report, then Download File → CSV):

Reports → Life cycle → Engagement → Pages and screens
Reports → Life cycle → Acquisition → Traffic acquisition
Reports → Life cycle → Engagement → Events

Set the date range to the last 30 days before you export.

Step 2: Upload and ask the big question. Start a new conversation with Claude or ChatGPT. Upload all three CSVs. Then paste something like this:

I run [a progressive nonprofit / a campaign / a small advocacy group] at [URL]. My main goals are [e.g., newsletter signups, event RSVPs, donations, calls to action]. I've attached the last 30 days of Google Analytics data. Tell me: (1) What's working that I should do more of? (2) What's underperforming and why? (3) What are three specific changes I should test next month?

Step 3: Follow up. Once you have the answer, keep going. Good follow-ups:

"Which pages convert visitors into [your goal]? Which don't?"
"Am I losing traffic on mobile vs. desktop? What might be causing it?"
"What content themes should I make more of based on what people are actually reading?"
"If I only had time to fix one thing on this site, what would have the biggest impact?"

Bonus: use AI to set GA4 up properly. If your GA4 isn't actually tracking the things you care about (signups, RSVPs, donations), take a screenshot of your GA4 admin panel and ask Claude to walk you through configuring events and conversions for your specific goals. It'll give you step-by-step instructions with the exact buttons to click.

AI isn't replacing your data analyst here. It's giving a busy organizer enough of an analyst to make better decisions this month. Progressive orgs don't need more tools. They need more capacity, and this is a free way to get some.

From our friends

Your org deserves its own AI. Not Big Tech's.

Change Agent is a private AI platform built for nonprofits, unions, and advocacy orgs. Your data stays yours, it plugs into tools you already use (Google Drive, Slack, ActBlue), and it handles the tedious stuff so your team can focus on the mission. Starts at $35/month. Small nonprofits under $1M can apply for discounted pricing.

Learn more

Looking Ahead

Tuesday, April 20: California Senate floor vote on SB 947, the No Robo Bosses Act. If you live in California, this is the week to make noise. If you don't, this is the week to start pitching your own state legislators on a companion bill. Progressive policy wins are built on templates. Someone has to go first. It might as well be California.

And if you're looking for a way to engage that doesn't involve another news cycle: go watch Sky Speirs. Then go send her playlist to someone who thinks AI is either going to save the world or end it. The real answer is harder and more interesting, and our side has the better version of that answer.