This website uses cookies

Read our Privacy policy and Terms of use for more information.

🦠 Cyclosporiasis update

📜 Anti-democracy amendment fails

💰 Why Trump’s fundraising matters

🤖 AI hacks

Hi friends,

Chris Vazquez, your Friday newsletter writer here. Another week has come and gone, and the virus making me too afraid to order my beloved Beefy Nacho Loaded Griller from Taco Bell is still here. But at least we have Dave videos, and a newsletter to break them down.

Before we dive in, it’s my last chance to remind you about our huge anniversary sale! You only have a few more days to get our best offer ever, 25% off annual subscriptions. You get all these perks for only a few dollars each week.

Now, let’s dive in!

What did Taco Bell know and when did they know it?

  • Lots of Taco Bell locations have removed Taylor Farms lettuce and other produce from some of their menu items because of the ongoing cyclosporiasis outbreak. Michigan health authorities were even in touch with Taco Bell’s parent company, the aptly named Yum! Brands Inc. $YUM ( ▼ 0.71% ) , about the outbreak weeks before the restaurant chain alerted customers about it.

  • So, am I panicking because I ate lettuce last night? Yes. Should I be? Maybe not! USDA data does show a higher-than-usual amount of recalls on meat, poultry and eggs. But if you’re a soy boy like me, there may be less to worry about — USDA and FDA data shows that overall food and beverage recalls so far this year are on pace with recent years.

  • It’s hard to answer if our food is getting less safe. Experts say the Trump administration’s federal funding cuts are slowing responses to things like this cyclosporiasis outbreak, and hurting food safety overall. At the same time, scientific advances in surveillance, lab testing and more have made us more prepared to respond to contaminations before they become huge problems.

  • Maybe that doesn’t help you if you’re deciding how scared to be of the produce in your fridge. But this website might. It’s the USDA’s tool for looking up recalls.

A MESSAGE FROM OUR SPONSOR

Dave here in this message on behalf of our partner beehiiv, the reason this newsletter even exists! beehiiv recently released a bunch of cool new features, including a visual editor and community features.

beehiiv is a fantastic platform for creators and independent journalists (like me)! Use my special code DAVE30 and get 30% off your first three months.

Dave here again (remember me?) after the beehiiv graphic to remind you (in case you’ve already forgotten) to use code DAVE30 and get 30% off your first three months of beehiiv.

Putting the “local” in “Local News International”

  • In Missouri, where Dave lives, more than 80% of voters rejected a Republican-led amendment. To explain what this amendment sought to do, let’s say Dave wants to pass a tax on Spam that would directly fund the production of whimsical yet informative news videos. Under the current system that’s been in place since 1908, Dave can collect signatures to put the Spam News Tax question on the ballot. If more than half of Missourians vote yes, it passes.

  • Amendment 4 would have changed that. It would have required these kinds of citizen initiatives to get a majority not just across the state, but in every single congressional district. And that would’ve made it way harder to get those initiatives on the ballot. In other words, it was an effort to limit direct democracy in Missouri.

  • Why does this matter, even if you don’t live in Missouri? The kind of citizen initiatives that Republicans were trying to limit have been used to legalize marijuana, protect abortion rights and expand Medicaid. Just slightly over half of all US states allow citizen-led ballot measures, and rolling back that right in Missouri could have been used as a template for other states. Instead, voters overwhelmingly rejected it. Now, will they reject the same people who tried to roll back this right when they’re up for election in November?

  • That brings us to this week’s media literacy tip from our friends at MediaWise! If you’re trying to figure out exactly who or what you’re voting on this election cycle, here are some things you can keep in mind.

Who is Trump’s “Princess of Darkness”?

  • Trump’s ballroom project, his future presidential library that may have a whole airplane, and his other pet projects all cost money to build. And where does that money come from? The Princess of Darkness, of course.

  • Trump reportedly gave his fundraiser Meredith O'Rourke that nickname, and she’s getting huge sums of money from companies. The Wall Street Journal reports that Trump’s personal control over fundraising and the amount he’s raising for projects like these are unusual.

  • The money’s not just going to construction projects. It’s also going to Trump’s super PAC, which had $400 million as of two weeks ago. And that super PAC hasn’t really spent money on candidates ahead of the midterms. Instead, POLITICO reports it’s spending on consulting, bank fees, credit card statements and more.

  • And Trump also happens to do exactly what his donors want a lot of the time. For instance, he met with tobacco executives and interrupted that meeting to call administration officials about easing vaping restrictions. Those restrictions ultimately were eased, surely to the delight of the teenagers I was surrounded by at Lollapalooza last weekend, and Big Tobacco also gave Trump millions of dollars.

The robots keep hacking people

  • So Meta made an AI model, and Meta hired a company called Irregular to test that AI model’s hacking capabilities. And boy, is it capable of hacking. The Wall Street Journal reports that the model made it onto the internet and hacked another company. Which model was it, who got hacked, and how long was this model on the internet for? We don’t know! Meta wouldn’t say. But we do know this wasn’t supposed to happen. Meta said in a statement that a “misconfiguration” on Irregular’s part “inadvertently allowed” the hacking.

  • Irregular also does testing for Anthropic, which also had AI models hack three other organizations during what was supposed to be an isolated test. Irregular said in a statement to Reuters that it “is developing a white paper to share best practices for containment and securely running cyber evaluations.” It remains to be seen whether a memo about why the robots are going rogue will stop the robots from going rogue.

  • OpenAI also recently had an AI model get onto the internet on its own and hack another company — albeit under different circumstances than the recent Meta and Anthropic incidents. So, this is a pattern. It happens when companies test their AI models under a set of guardrails. But AI has either found ways around those guardrails or found ways to get online due to errors in the test itself. But researchers say that in all these cases, it’s not like AI is gaining consciousness and deciding to go online; rather, this is all happening under the oversight of real people in charge of AI.

Each week, after running through the news Dave has covered in the past week, I turn things over to him for some analysis. Dave, take it away!

this guy might be the best on YT

@plasticwrapcharlie

I am very wary of making content about the YouTube drama surrounding Hank Green and AI. I don’t want to talk about it for engagement. I want to discuss it because I think it’s a really interesting conversation, one that I think Colin and Samir handled pretty well in their most recent video.

So, for now, I’ve settled on talking about it here in my cozy corner. For those who aren’t aware, here’s what happened, with a quick refresher on Hank Green, too:

YouTube legend Hank Green is one half of the Vlogbrothers. It’s a channel he’s run with his brother John for two decades (the very same John who made you cry while reading and later watching The Fault in Our Stars and Turtles All the Way Down). Hank is incredibly smart and has built a reputation on his thoughtfulness and intellect. I was thrilled to be a part of his Study Hall series a couple years back. 

Hank admitted to using AI while researching for videos, and further admitted that he has leaned on it more to deal with his massive workload. Then he deleted that tweet (which, like the word “twitter,” obviously lives forever in screenshot form).
Importantly, here’s what he has said since, from a much lengthier Reddit post:

  1. He is not and was not using it to write scripts.

  2. He felt that AI was giving him a dopamine hit that was “not healthy.”

  3. He is pausing videos on Hankschannel for now, and will possibly come back with hired writing and research support (humans, I presume).

  4. He is mortified to have let so many people down.

Looking at his Reddit post as objectively as I’m able, I would say it’s a pretty thoughtful, human response to an admission of AI overuse. And I think instead of piling on to someone who clearly is trying to navigate this graciously, we should take this moment to discuss the use of AI and how it should be disclosed going forward.

I don’t think that all AI is inevitable, but I do think that some versions of AI have been irreversibly integrated into systems that we literally can’t avoid. Here’s an example:

At LNI, we don’t use any generative AI to make our videos and I’m very proud of that. I read as much as I can from different sources, write a script, shoot it all myself, and edit it in Premiere Pro. A human (our co-founder, Lauren) does a couple, sometimes several, passes on it before it’s posted. I love the creative process and what comes as a result. Most of all, I love watching other humans interact with what I created from scratch.

But AI isn’t technically omitted from that entire process. For example, in yesterday’s video about the Missouri citizen-led initiative proposal, I was trying to find a good song to play underneath the silent film bit. I went to APM Music, a site we pay for to license (human-made) music.

Here’s the catch: APM recently introduced AI into their search engine. So, when I typed “early 1900s silent film background” into the search bar, I was technically using AI to find the song.

Again, I don’t think all of AI is inevitable, but it’s not entirely avoidable. Even right now, if you’re reading this from your smartphone, it’s very likely using AI in the background, even if you turned off “Apple Intelligence” (watch this video to find out how to turn that off). Predictive text and the original Siri, for example, are early versions of AI.

For a few months now, I’ve been proposing a scale of AI use. If a “10” is AI slop, then a “1” is using a search bar you can’t avoid. In general, I think we need to create more comprehensive disclosure language around when we’re using AI and how we’re using it and set those standards for our viewers. The government isn’t doing much to regulate AI, so maybe it’s time the rest of us start to self-regulate and create this system. 

Personally? I don’t want to watch anything that used any form of generative AI. I’m also not super stoked about using AI to pull research. I like knowing it was created and carefully studied by a human. Other people don’t mind that. Fine by me. I think it should be about that transparency between creator and viewer, so no one has to guess anymore.

If you made it to the end of this newsletter, you get two rewards. The first is a pet picture from a loyal reader. This is Brownie!

Me after reading the news

Your second reward is the reveal for this week’s link scavenger hunt, in which I hid a non-news related link somewhere in the body of this email. The last link in the section about recent AI hacks goes to a summary of an X-Men story I just finished reading about Professor X’s mutant-tracking computer gaining sentience. Those comics were, to be honest, not good! So I’m glad they’re not our reality just yet!

Until next week!

Chris

Reply

Avatar

or to participate

Keep Reading