GD POLITICS

A dam of pent-up concern over AI safety burst this past week. It started with a warning from AI researcher Jacob Coxon, who resigned from Anthropic and wrote on X: “Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources.”
He continued: “The people building AI earnestly believe that it could kill us all by the end of the decade.”
Since then, Anthropic CEO Dario Amodei has outlined a plan to slow the technology’s development and give safety measures time to catch up. He also warned that, within six months to a year, swarms of AI agents could take over the entire internet. Sam Altman and Elon Musk both expressed support for Amodei’s warning and proposed approach.
Politicians are also taking notice and calling for regulation. The most forceful proposals have come from Democrats, though not exclusively. Sens. John Thune, Amy Klobuchar and Ted Cruz are reportedly working on AI-safety legislation that could be introduced as soon as this week.
Bernie Sanders has called on President Trump to pursue an AI agreement with China modeled on the nuclear arms agreements reached by Ronald Reagan and Mikhail Gorbachev. At a fundraiser, former President Barack Obama urged Democrats to “aggressively develop political and governing agendas to prioritize the issue in the coming years,” according to The New York Times.
For his part, Trump wrote Monday on Truth Social: “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT.” He continued: “There is a SICK conspiracy going on against AI and Data Centers, and the only one that is happy about it is China. WHOEVER WINS AI, WINS!”
Joining me today to discuss whether AI is going to kill us all — or, at the very least, how the threat that it could is ricocheting through American politics — are two sharp thinkers on AI. David Byler is a pollster and vice president at National Research Group. He writes the newsletter “Real Research on Artificial Intelligence.” Mary Radcliffe is head of research at FiftyPlusOne.


This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.gdpolitics.com/subscribe

This is a free preview of a paid episode. To hear more, visit www.gdpolitics.com

As you’ve likely heard, Nate Silver, Clare Malone and I have a new podcast going over at Crooked Media called Still Counting. We released our first official episode on Wednesday, and that is what you’re going to hear in the feed today.
(It’s this thing we do in the podcasting world called a “feed drop.” The hope is that you will like what you hear and subscribe to the new podcast. So please, dear listener, go subscribe!)
To pull the curtain back a little, starting this new podcast is something Nate, Clare and I have been talking about doing since around February or March of this year. We really enjoyed taping live shows at the Comedy Cellar, and audience members would frequently ask us after the shows if we’d consider getting back together.
We all have our own projects going – Clare at the New Yorker, Nate at Silver Bulletin, and me here at GD – but the stars aligned just so that it felt like reassembling the band made sense.
We got to work reaching out to different podcast shops that we thought might be interested in a FiveThirtyEight sequel, and we had a lot of conversations. One of the things that attracted us to Crooked was that they were familiar with FiveThirtyEight, knew what we were about, and wanted to let us do our thing. That included not aligning with a political party.
We set a goal to launch as election season was picking up after Labor Day, and voilà, here we are.
In this first episode, we dive into the most competitive Senate races in the midterms according to the polls: Texas, Michigan, Iowa, Maine, Alaska, North Carolina, New Hampshire, Kansas, and Ohio. We also weigh whether the Republican midterm convention is a “good use of convention or a bad use of convention.” And, on a more somber note, we take stock of the year in American politics since Charlie Kirk’s assassination.
It was nice to get the crew back together, and we’re still just getting warmed up. So, once again, make sure to subscribe to join us on our new adventure!

Quick note: The mid-decade redistricting wars have been so chaotic that the maps changed again between when I recorded this podcast on Thursday and when you’re hearing it.
Hours after I spoke with Dave Wasserman of the Cook Political Report, the Missouri Supreme Court blocked Republicans’ new congressional map from being used this November and ordered the state to move forward with a referendum on it. Missouri is appealing to the U.S. Supreme Court, but if the decision stands, Republicans will have one fewer potential pickup from redistricting this cycle. Consider that an asterisk on some of the numbers you’ll hear in this episode.
The bigger story remains the same: The fight over congressional maps has transformed the battle for the House. Ten states have redrawn their maps since 2024, and according to Cook’s Partisan Voter Index, the median House district has moved from roughly one point more Republican than the country to about three points more Republican.
At the same time, competitive districts are disappearing. In 2000, 164 House districts fell between R+5 and D+5 in Cook’s PVI. Today, Wasserman says there are just 78 — fewer than one in five House districts. And the redistricting wars aren’t necessarily over. Both parties could still try to squeeze additional seats out of states they control in the years ahead.
Dave and I first dug into gerrymandering together nearly a decade ago for FiveThirtyEight’s Gerrymandering Project. Back then, redistricting was largely something that happened once every ten years. As Dave puts it today, we’ve gone from redistricting every ten years to what feels like “every 10 seconds.”


This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.gdpolitics.com/subscribe

This is a free preview of a paid episode. To hear more, visit www.gdpolitics.com

The full episode is available to paid subscribers. Once you become a paid subscriber, you can connect your account to your preferred podcast player here.
We are exactly two months away from Election Day and just days away from the official start of election season. It’s going to be two months of new polls, forecast gyrations and, in a word, uncertainty.
At this point in an election cycle, back at FiveThirtyEight, I’d sometimes feel that a therapist might be better suited to hosting the podcast, because the job became, to some extent, helping people reckon with the unknown.
One way we did that was by trying to quantify uncertainty and put it in historical context — looking at, for example, the average error or the largest error in past elections. And I think that’s plenty useful.
But after more than a decade in the data journalism trenches, I can tell you that numbers will soothe the soul only so much. At a certain point, we just have to get comfortable with not knowing.
And that’s what we’re doing on today’s podcast, in the final calm before the storm. My guest is journalist Simone Stolzoff, author of How to Not Know: The Value of Uncertainty in a World That Demands Answers.

We’re a week away from the official start of election season — Labor Day — but the polls are not waiting. New Senate surveys in Texas, Michigan and Iowa are sharpening the picture of the battle for Congress. And polls from South Carolina, Nebraska and Kansas have made Democrats optimistic about competitive races on decidedly red turf.
So is this silly season, when Democrats get their hopes up in states they won’t actually win? Or is a blue wave putting much more of the map in play? Lakshya Jain, founder of Split Ticket and head of political data at The Argument, and Zach Donnini, head of data science at VoteHub, join today’s episode to sort through the evidence.
We discuss why Texas looks different from Democrats’ previous attempts to turn it blue, how much candidate quality could matter in Michigan, and why a favorable national environment likely isn’t enough for Democrats in South Carolina.
We also check in on Zach and Lakshya’s forecasts and the Democratic primary between Ed Markey and Seth Moulton in Massachusetts. (Yes, we are still talking about primaries. Those don’t wrap up until mid-September!)


This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.gdpolitics.com/subscribe