Glowing Red Eyes and the First Amendment: Oregon's Deepfake Disclosure Law Faces Its First Real Test - August 15, 2026

Glowing Red Eyes and the First Amendment: Oregon's Deepfake Disclosure Law Faces Its First Real Test - August 15, 2026 Oregon's Secretary of State is now investigating two absurd AI generated attack ads under a state disclosure law, and the case has become the first real stress test of whether roughly thirty state deepfake statutes can stop deception without trampling the First Amendment. Chris and Laura dig into the outlandishness defense, the courts that have already struck similar laws down, the enforcement timing gap, and why provenance may end up mattering more than detection. Hosted by Chris and Laura. The DX Today Podcast brings you daily deep dives into the most consequential stories in the AI ecosystem. #AIDeepfakes #ElectionLaw #FirstAmendment #AIRegulation #SyntheticMedia

Transcript

SPEAKER_01 0:00

Welcome to the DX Today Podcast, your daily deep dive into the AI ecosystem. I'm Chris, and joining me as always is Laura.

SPEAKER_00 0:07

Thanks, Chris. I am genuinely excited about today's story because it sits right at the collision point between generative AI, campaign politics, and the First Amendment. And it is finally getting tested for real.

SPEAKER_01 0:19

So set the stage for us, because I keep hearing that this is the first midterm cycle where AI-generated political ads are not a hypothetical anymore, but an actual thing regulators are chasing in real time.

SPEAKER_00 0:31

That's exactly right. And the clearest example comes out of Oregon, where a law called Senate Bill 1571, enacted back in 2024, requires candidates to disclose when they use AI to create their campaign ads.

SPEAKER_01 0:44

Okay, a disclosure law rather than an outright ban. That distinction feels important. So before we get into the specific case, walk me through what disclosure actually means in practice for a campaign running one of these ads.

SPEAKER_00 0:57

In practice, it means the ad has to carry a clear label telling voters that synthetic or AI-generated media was used. So the idea is not to censor the message, but to make sure nobody is fooled about how it was made.

SPEAKER_01 1:09

And now there is a live test case putting that theory under real pressure. Tell me about the candidate and the ads that apparently kicked off an actual investigation by the state. Because the details here are genuinely wild.

SPEAKER_00 1:22

So the candidate is Jonathan Lockwood, who ran unsuccessfully in 2024 in Oregon's fifth congressional district primary. And he produced two AI-generated ads targeting Representative Janelle Bynum that are frankly hard to describe with a straight face.

SPEAKER_01 1:37

I have a feeling I know where this is going, but go ahead and describe them. Because I think our listeners need to understand just how surreal AI-generated political content has already become in a real American election.

SPEAKER_00 1:48

One ad depicted Representative Bynum and Governor Tina Kotec with glowing red eyes chasing frightened children down a street. And a second showed Bynum in a McDonald's uniform placed into a series of completely fabricated scenarios.

SPEAKER_01 2:01

Red glowing eyes chasing children. I mean, that is so cartoonishly over the top that it raises the exact question the candidate himself is now using as his legal defense, which we should get into right now.

SPEAKER_00 2:13

Right, because Lockwood's argument is essentially that the ads were so outlandish that no reasonable person could ever mistake them for reality. And therefore, in his view, no disclosure was needed because nobody was actually being deceived.

SPEAKER_01 2:26

And that is a surprisingly serious argument, not just a dodge, because satire and parody sit at the absolute core of protective political speech in this country. So how is Oregon actually responding to that complaint so far?

SPEAKER_00 2:37

In July of 2026, the Oregon Secretary of State's office launched an investigation after receiving a formal complaint. And if the office finds a violation, the law allows for prosecution and fines reaching up to $10,000.

SPEAKER_01 2:51

$10,000 is not nothing for a small campaign, but I want to zoom out because Oregon is clearly not alone here. How widespread is this kind of law across the country heading into these midterms?

SPEAKER_00 3:01

It is surprisingly widespread now, with roughly 30 states carrying some form of election-related AI deepfake law on the books. And Maryland was one of the most recent states to join that growing list of jurisdictions.

SPEAKER_01 3:14

30 states is a lot faster than I would have expected government to move on anything involving technology. So what does the typical version of one of these laws actually look like when you strip it down?

SPEAKER_00 3:23

Most of them follow the same basic template, dictating when disclosure of AI use is required, and giving candidates who get targeted by noncompliant synthetic ads a way to seek recourse directly through the courts.

SPEAKER_01 3:36

So it is mostly a disclosure and private lawsuit model rather than a criminal crackdown, which makes sense given the constitutional minefield. But not every one of these laws has survived contact with a judge, has it?

SPEAKER_00 3:47

No, and this is where it gets legally messy because deep fake election laws in both California and Hawaii were struck down on First Amendment grounds, which tells you these statutes are walking a genuinely narrow constitutional line.

SPEAKER_01 4:00

That is the crux of the whole thing for me. Explain why a well-intentioned law aimed at stopping deceptive fakes can still run headfirst into the First Amendment and lose because that surprises a lot of people.

SPEAKER_00 4:12

The core problem is that political speech gets the very highest level of constitutional protection. And courts are deeply skeptical of any rule that could be used to punish satire, exaggeration, or unpopular opinions dressed up as an election message.

SPEAKER_01 4:26

So a law has to be narrow enough to only catch genuinely deceptive fakes, while somehow not sweeping in the cartoon with the glowing red eyes, which a candidate can plausibly call obvious political satire. That is a brutal needle to thread.

SPEAKER_00 4:40

It really is, and that is precisely why the Lockwood case is so instructive. Because it lands right on that fault line between deceptive impersonation, which the state can likely regulate, and absurd caricature, which is much harder to touch.

SPEAKER_01 4:53

Let me play devil's advocate for a second, because I can hear some listeners thinking the whole disclosure approach is basically toothless. If the penalty is a fine that arrives months later, does the label even matter during a fast campaign?

SPEAKER_00 5:05

That is a fair and important challenge because the timing problem is real, since an investigation and a fine can easily land long after the votes are counted. By which point, whatever damage the ad did to a race is already done.

SPEAKER_01 5:17

So enforcement speed is almost as important as the rule itself. And a law that only bites after election day is arguably solving the wrong half of the problem. Is anyone actually designing around that timing gap?

SPEAKER_00 5:28

Some states are trying by pairing disclosure rules with faster injunctive relief, meaning a targeted candidate can ask a court to order the ad taken down or labeled immediately rather than waiting for a slow penalty process to grind along.

SPEAKER_01 5:42

That makes sense, but injunctions against political speech are exactly the kind of thing appeals courts get nervous about. So I imagine that fast takedown power is also going to draw its own First Amendment challenges pretty quickly.

SPEAKER_00 5:54

Almost certainly, and you can already see the shape of the fight, because a rushed order pulling down a political ad right before an election is arguably a bigger free speech intrusion than a modest fine imposed quietly afterward.

SPEAKER_01 6:06

Let's talk about the federal picture, because so far everything we have discussed is a patchwork of individual states doing their own thing. Is there any coherent national standard coming to tie all of this together?

SPEAKER_00 6:17

Not yet. And that is a real gap because there is currently no federal standard governing AI and election ads, which leaves us with a genuine patchwork where an identical ad could be perfectly legal in one state and prosecutable in the next.

SPEAKER_01 6:31

A patchwork like that seems almost designed to create confusion for national campaigns that run identical ads across dozens of media markets. Is there any movement in Congress to actually replace it with one unified rule?

SPEAKER_00 6:43

There is some, at least in signal form, since House Democrats have indicated they would introduce federal legislation on synthetic election media if they gain power. Though whether that ever becomes an actual enforceable law is very much an open question.

SPEAKER_01 6:57

So the honest summary is that the states are sprinting ahead while the federal government is still deciding whether to lace up its shoes. Who is doing the loudest advocacy work pushing all of this activity forward?

SPEAKER_00 7:09

A big driver is the advocacy group Public Citizen, and their democracy organizing manager, Alana Beller, framed it by saying state lawmakers continue to lead the charge in protecting constituents from the threat of generative AI deceiving voters.

SPEAKER_01 7:22

And to be clear, this is not purely theoretical or limited to fringe primary candidates. We are already seeing this kind of AI content from serious contenders and high-profile races this cycle, aren't we?

SPEAKER_00 7:33

Absolutely. And one cited example is U.S. Senate candidate Mike Rogers, a Michigan Republican, who posted AI campaign content on the platform X. And the reporting notes that some of this material was labeled while other pieces were not.

SPEAKER_01 7:48

That mix of labeled and unlabeled content from a major candidate really captures the enforcement nightmare. Because it means the same campaign can be compliant in one post and potentially in violation in the very next one they publish.

SPEAKER_00 8:00

Exactly. And it also shows that the norms are still being negotiated in real time. Because when even sophisticated campaigns are inconsistent about labeling, it signals that neither the rules nor the expectations have fully settled into place yet.

SPEAKER_01 8:14

Before we zoom out, I want to touch the technology layer. Because a lot of people assume we can simply build a detector that automatically flags every AI-generated ad. Is reliable automated detection actually realistic right now?

SPEAKER_00 8:27

Honestly, detection after the fact is a losing arms race. Because every improvement in spotting fakes gets fed straight back into training the next generation to be harder to catch, which is why many experts now favor provenance over detection.

SPEAKER_01 8:41

Provenance meaning you attach a verifiable record of how a piece of media was made at the moment of creation, rather than trying to reverse engineer its authenticity later. So where exactly do the big platforms fit into that whole approach?

SPEAKER_00 8:54

The platforms are absolutely pivotal because a disclosure law only works if the place where an ad actually spreads is willing to surface those labels prominently. And right now, enforcement leans heavily on companies whose incentives do not always favor slowing content down.

SPEAKER_01 9:10

Let me pull us up to the big picture because underneath the legal mechanics, there is a deeper question about trust. Are these laws really about the fakes themselves or about something bigger happening to the whole information environment?

SPEAKER_00 9:21

I think it is genuinely about the bigger thing, because the deepest danger is not just any single convincing fake, but the erosion of a shared baseline of reality where voters can no longer assume that what they see actually happened.

SPEAKER_01 9:34

That is that is the liar's dividend problem, right? Once people know convincing fakes exist, a politician caught on real video can simply claim it was AI generated and suddenly disclosure laws are fighting a battle on two fronts at once.

SPEAKER_00 9:47

That is precisely it, and it is why disclosure alone can never be the whole answer. Because a label helps you flag the fakes, but it does nothing to stop a bad actor from falsely branding authentic footage as a fabricated deep fake.

SPEAKER_01 10:00

So if you're a state lawmaker right now, you're essentially trying to build a system that both catches deception and protects satire and preserves authentic evidence, all while staying inside strict constitutional limits. That is a genuinely hard design problem.

SPEAKER_00 10:15

It is one of the hardest policy design problems out there. And honestly, the Oregon case is valuable precisely because it forces courts to actually draw those lines in a concrete fact pattern rather than in the abstract.

SPEAKER_01 10:27

Before we wrap, give me your honest read. When this Oregon investigation resolves, do you expect it to clarify where that line sits? Or do you expect it to just add another layer of legal uncertainty on top?

SPEAKER_00 10:38

My honest read is that it will clarify one narrow but crucial point, which is whether the outlandishness of an ad can function as a legal shield. And that single answer will ripple outward to shape how every other state writes and enforces these rules.

SPEAKER_01 10:52

And I suspect that whatever Oregon decides, it will get appealed, tested, and argued over well past this midterm cycle, which means we are really watching the opening chapter of a much longer constitutional story unfold in slow motion.

SPEAKER_00 11:05

I completely agree. And I would just add that the technology is not going to politely wait for the law to catch up. So every election from here forward is going to be a live experiment in exactly this tension we have been unpacking.

SPEAKER_01 11:17

That is a perfect note to land on, because it reframes all of this not as a solved problem, but as an ongoing negotiation between what AI can generate and what our democratic institutions are willing to tolerate. Any final thought, Laura?

SPEAKER_00 11:30

Just that voters themselves are part of this system too, and building the habit of pausing before you believe or share a striking political clip may end up mattering just as much as any statute a legislature manages to pass.

SPEAKER_01 11:42

Beautifully said, and a genuinely useful takeaway for everyone listening as we head deeper into this election season. Thank you, Laura, for walking us through such a tangled and important story with so much clarity today. That's all for today's episode of the DX Today Podcast. Thanks for listening, and we'll see you next time.