SThe Impact

Your weekly look at the intersection of AI, advocacy, and politics from the team at MFStrategies | www.MFStrategies.com

This Week’s Toplines

The Trump FTC is now asserting the power to judge whether AI outputs are accurate, and it's using that authority to steamroll state-level AI laws along the way.

At the same time, governors from both parties are scrambling to regulate campaign deepfakes, and California is staring down a choice between putting guardrails on its AI economy or letting it run loose. Then there's the nonprofit sector, where algorithms are quietly being used to deny people food and shelter with zero appeals process, all in a policy gap that the federal government has simply walked away from.

What seemed like bipartisan agreement on oversight is splintering into turf wars over who actually gets to define regulation: Is it Washington appointees enforcing ideological conformity? States writing a patchwork of conflicting disclosure rules? Or nobody?

The real question at this point isn't whether AI gets regulated. It's whether whatever regulation does show up is designed to protect people or to consolidate power.

News of the Week

The Oklahoman
Takeaway
At the National Governors Association summer meeting on July 30, Oklahoma Governor Kevin Stitt pushed for state-level laws on AI in political campaigns, mentioning he was considering calling a special session to address it. He pointed out that 31 states have already passed such laws. The following day, the NGA announced a $1 million partnership with RAISE US aimed at helping states develop AI policy frameworks. Maryland Governor Wes Moore, who is set to take over as NGA chairman, noted that his state already has laws on the books that prevent AI from being used in campaigns.
Why it matters
Governors from both parties are pushing hard on AI deepfake regulation, and that pressure could lead to a patchwork of state disclosure and labeling laws well before Congress gets around to acting. For campaigns using synthetic media, this creates real compliance headaches, since vendors will have to tweak their processes to match whatever each state requires. There will almost certainly be legal battles over where the line falls between something misleading and something that's clearly satire. And consultants who put out AI-generated content without proper labels? They're going to find themselves on the hook in ways they haven't had to worry about before.
 
Electronic Frontier Foundation
Takeaway
In July, the FTC put out a proposed policy that would give the agency power to judge whether AI outputs are accurate. The proposal claims authority to override state AI laws and threatens companies with enforcement action if they don't comply. EFF and civil rights groups responded with joint comments calling for the proposal to be withdrawn entirely. Their argument: it violates the First Amendment by turning the government into the arbiter of truth, it goes beyond what the FTC is legally allowed to do when it comes to preempting state law, and it amounts to a vague jawboning scheme designed to pressure AI developers into falling in line with the Trump administration's ideology. The whole thing traces back to Trump's "Preventing Woke AI" executive order.
Why it matters
The FTC is now asserting the authority to evaluate AI accuracy and preempt state-level AI laws. In practice, this opens a path toward viewpoint-driven enforcement shaped by White House ideology. It also hands the administration a tool to push companies toward self-censoring their AI outputs, all while blocking state regulations that address bias in hiring, housing, or healthcare. For campaigns and advocacy groups, the result is a fog of uncertainty around which messages or tools might invite federal scrutiny.
 

Ipsos
Takeaway
Recent Ipsos polling finds that Republicans and Democrats actually agree on something: the government should be regulating AI. Majorities in both parties want oversight when it comes to economic stability and public safety, and voters on both sides seem more comfortable with guardrails than with Silicon Valley's "move fast and break things" mentality. That said, the consensus fractures on specifics like anti-discrimination enforcement and environmental concerns. One thing pollsters are watching closely: if AI starts wiping out jobs at companies people already distrust, the political fallout could be fierce, and it won't be limited to one party.
Why it matters
Bipartisan support for AI regulation is one of the few areas where both parties actually agree, but that agreement hands real leverage to whoever gets to define what 'oversight' looks like in practice. If job losses start piling up, populist lawmakers on both sides will almost certainly use worker anger against tech companies and their executives. Campaigns need to keep an eye on discrimination and environmental concerns in particular. Those are the issues most likely to blow up the current consensus and put candidates in a position where they have to pick a lane.
 
StateScoop
Takeaway
California's governor's race this November comes down to a pretty simple question: who gets to set the rules for AI, cybersecurity spending, and government tech for 40 million people living in one of the world's biggest tech economies? Democrat Xavier Becerra is pushing for tougher AI safety laws, stronger worker protections, and publicly funded AI infrastructure through CalCompute. Republican Steve Hilton wants to go the other direction, building a state-level version of the federal Department of Government Efficiency to slash regulations and speed up permitting. They actually agree on one thing: using AI to improve state services like permitting and benefits delivery. Where they split is on the bigger question of whether government should be actively regulating AI or stepping back and letting the market figure it out.
Why it matters
Whoever wins the California governor's race will have a lot of say over AI procurement, safety enforcement, and data center permitting. Both candidates want to expand AI use, but they're far apart on how much regulation should come with it. That gap matters for state vendors trying to plan ahead, for agencies working against compliance deadlines, and for local governments that have been using California's approach as a template for their own AI rules. There's also a bigger political question here: campaigns are paying attention to whether tech policy moves votes, and this race could show whether people care more about moving fast on innovation or holding algorithms to account.
 
The Fulcrum
Takeaway
A growing number of nonprofits in the U.S. have started relying on AI algorithms to make calls about who qualifies for food assistance, emergency shelter, and medical referrals. The problem is that these organizations aren't bound by the same legal protections, due process requirements, or oversight mechanisms that govern welfare decisions made by government agencies. Studies have found that the algorithms in question can replicate existing inequalities and end up turning away the people who are most in need of help. Worse, there's no formal appeals process for someone who gets denied. And right now, federal AI policy doesn't cover nonprofit welfare delivery at all, which means algorithmic decision-making in social services is running with almost no accountability.
Why it matters
The original text used an em dash, had a clipped and directive tone ('Expect compliance pressure'), and relied on sentence fragments that could read as AI-generated shorthand. The rewrite removes the em dash, expands fragments into full sentences, uses natural transitions ('On top of that,' 'which means'), and adopts a conversational but informative register that reads like a human newsletter writer explaining a policy issue to their audience.
 

Worth Thinking About This Week

"The government may not install itself as the arbiter of truth." -EFF, Public Knowledge, and Fight for the Future