Latest / Question Everything / A Video So Real It Looks Fake

Transcript

A Video So Real It Looks Fake · Question Everything

• 0:01
  Speaker 1: Hey, everyone. Today, I'm going to sit down with somebody
• 0:08
  Speaker 1: who spends all day tracking lies on the Internet. And together,
• 0:12
  Speaker 1: she and I are going to dissect a single lie
• 0:14
  Speaker 1: that's been rocketing around the world this week, getting millions
• 0:17
  Speaker 1: and millions of views on social media, affecting the way
• 0:20
  Speaker 1: people think about the war in Iran, being repeated by
• 0:22
  Speaker 1: state media. And now the lie is being parroted by
• 0:25
  Speaker 1: the biggest podcaster in the world. Yours truly. Just kidding.
• 0:30
  Speaker 1: It's Joe Rogan. I spend a lot of time thinking
• 0:33
  Speaker 1: about lies myself. But I gotta say, digging into this
• 0:36
  Speaker 1: one example, taking a single lie and putting it under
• 0:40
  Speaker 1: the microscope, it was illuminating to see all the facets
• 0:43
  Speaker 1: of it and understand better why people might believe it.
• 0:47
  Speaker 1: And then when you imagine that multiplied many, many times
• 0:50
  Speaker 1: over with all the lies people are exposed to every day,
• 0:53
  Speaker 1: it was like seeing the world, the whole information war
• 0:56
  Speaker 1: around America and Israel's assault on Iran, in a grain
• 0:59
  Speaker 1: of sand.
• 1:01
  Speaker 2: We were fact-checking so much AI-generated content. It was unlike
• 1:06
  Speaker 2: something I've seen during my time at this company, just
• 1:10
  Speaker 2: the sheer volume of fake videos and photos that were
• 1:13
  Speaker 2: being spread online.
• 1:14
  Speaker 1: Wow, so you feel like the last three weeks, this
• 1:17
  Speaker 1: conflict is a new theater of war, basically, like you're
• 1:20
  Speaker 1: on the front lines of something new right now. We
• 1:22
  Speaker 1: all are.
• 1:23
  Speaker 3: I definitely think so.
• 1:25
  Speaker 1: From Placement Theory and KCRW, This is Question Everything. I'm
• 1:29
  Speaker 1: Brian Reed. Stick around. Sophia Rubinson is a senior editor
• 1:40
  Speaker 1: at NewsGuard, a company that tries to make sense of
• 1:42
  Speaker 1: what's true and what's not online. Part of what they
• 1:45
  Speaker 1: do is rate the reliability of other outlets, kind of
• 1:49
  Speaker 1: like a nutrition label for news sites. Shout out KCRW
• 1:52
  Speaker 1: with 100% NewsGuard rating. NewsGuard also tracks disinformation and lies
• 1:57
  Speaker 1: online in real time to see which ones are taking off,
• 2:00
  Speaker 1: gaining traction, and which could be the most damaging. In
• 2:05
  Speaker 1: the last 10 months, Sofia and her team have started
• 2:08
  Speaker 1: publishing the false claim of the week. This is a
• 2:10
  Speaker 1: claim that their tools and analysis show is going extremely
• 2:14
  Speaker 1: viral that week, getting lots of engagement, and also has
• 2:17
  Speaker 1: a high potential to cause harm, maybe shift public opinion
• 2:21
  Speaker 1: or influence the actions of world leaders. Sophia and her
• 2:24
  Speaker 1: colleagues released the false claim of the week in a
• 2:25
  Speaker 1: newsletter each Friday called Reality Check. And I wanted to
• 2:29
  Speaker 1: talk to her about the latest one. Sophia, what is
• 2:32
  Speaker 1: the false claim of the week?
• 2:34
  Speaker 2: Right now, we're in the middle of this conflict in Iran.
• 2:37
  Speaker 2: You know, for the last three weeks, all of our
• 2:40
  Speaker 2: false claims of the week have been related to the
• 2:41
  Speaker 2: war in some way.
• 2:42
  Speaker 3: And this week was no exception. For the last week
• 2:45
  Speaker 3: and a half.
• 2:46
  Speaker 2: We've been seeing a lot of baseless claims in Iranian
• 2:49
  Speaker 2: state media claiming that Benjamin Netanyahu, the prime minister, of
• 2:54
  Speaker 2: Israel was either gravely injured or killed in an Iranian
• 2:57
  Speaker 2: missile strike.
• 2:59
  Speaker 3: Then, sort of in response.
• 3:00
  Speaker 2: To this, Netanyahu posted a somewhat satirical video on his
• 3:05
  Speaker 2: ex-profile where he shows that he has five fingers, he
• 3:08
  Speaker 2: shows that he's alive. He makes a couple of jokes
• 3:11
  Speaker 2: at a coffee shop.
• 3:12
  Speaker 1: It's a proof of life video, basically.
• 3:14
  Speaker 3: It's a proof of life, yes.
• 3:16
  Speaker 1: From himself.
• 3:17
  Speaker 2: From himself, posted on his own account. But in response,
• 3:20
  Speaker 2: we saw a lot of pro-Iranian.
• 3:23
  Speaker 3: Social media users.
• 3:25
  Speaker 2: We also saw this in Iranian state media claiming that
• 3:28
  Speaker 2: this video was AI, when in fact this was an
• 3:31
  Speaker 2: authentic video that he posted on his own ex-profile.
• 3:34
  Speaker 1: Okay, so I just want to break this down because
• 3:36
  Speaker 1: there's like a few layers going on here. So there's
• 3:39
  Speaker 1: one set of lies that have been going around claiming
• 3:42
  Speaker 1: that Netanyahu is dead, that he was killed in an
• 3:45
  Speaker 1: Iranian missile strike. He tried to disprove that by posting
• 3:49
  Speaker 1: a video of him being very much alive and drinking coffee.
• 3:53
  Speaker 1: And the false claim of the week is people online
• 3:56
  Speaker 1: and Iranian state media sources claiming that that video of
• 4:00
  Speaker 1: Netanyahu drinking coffee was AI generated. It was a deepfake. Yes.
• 4:04
  Speaker 1: Can we pull it up? This is it, right?
• 4:06
  Speaker 2: Yeah.
• 4:06
  Speaker 4: All right.
• 4:06
  Speaker 1: So this was posted March 15th. He posted this video
• 4:11
  Speaker 1: and his caption on it or his post on it was,
• 4:13
  Speaker 1: they say I'm what? Watch. And then this is what
• 4:16
  Speaker 1: happens in the video.
• 4:18
  Speaker 2: Mashallah tohru.
• 4:19
  Speaker 1: If you want to watch this video for yourself and
• 4:21
  Speaker 1: follow along as Sophia and I examine it, there's a
• 4:23
  Speaker 1: link to it in the show notes. Netanyahu's in a jacket.
• 4:26
  Speaker 1: He's at the counter of what looks to be a
• 4:28
  Speaker 1: coffee shop. His left hand is in his jacket pocket.
• 4:31
  Speaker 1: With his right hand, he picks up a cup of
• 4:33
  Speaker 1: coffee off the counter. Maybe it's a latte. There's a
• 4:36
  Speaker 1: little foam art on top. Someone off camera says to
• 4:40
  Speaker 1: him in Hebrew, they're saying online that you're dead.
• 4:43
  Speaker 3: And Netanyahu says cheekily, I'm dying.
• 4:49
  Speaker 1: For coffee, he pulls his hand out of his pocket.
• 4:52
  Speaker 1: And then what's he doing here exactly?
• 5:00
  Speaker 2: So here he's holding up his hand to show that
• 5:02
  Speaker 2: he has five fingers. So this is a reference to
• 5:05
  Speaker 2: another false claim, very similar, that was spreading online a
• 5:08
  Speaker 2: few days before he made this video. It was a
• 5:10
  Speaker 2: video of a press conference speaking to reporters about updates
• 5:15
  Speaker 2: from the war. And people were claiming that he had
• 5:17
  Speaker 2: a sixth finger. but it was really just his palm
• 5:20
  Speaker 2: of his hand. So they were using that as proof
• 5:23
  Speaker 2: that the press conference was AI generated when it's very
• 5:26
  Speaker 2: clear when you slow down the video that there is
• 5:28
  Speaker 2: no sixth finger.
• 5:30
  Speaker 1: Okay. So here he is. He's at this coffee shop.
• 5:33
  Speaker 1: He holds up both hands. Shows five fingers. And then
• 5:37
  Speaker 1: he has a sip of his coffee. And then he
• 5:40
  Speaker 1: goes on to say some kind of jingoistic stuff about
• 5:43
  Speaker 1: the IDF and stuff. So what were people saying about
• 5:46
  Speaker 1: this video? What did you see online?
• 5:48
  Speaker 2: There were several pieces of evidence that people were pointing
• 5:52
  Speaker 2: to as proof that this is supposedly AI. One is
• 5:55
  Speaker 2: that the level of the coffee doesn't drastically decrease when
• 5:59
  Speaker 2: he takes a sip.
• 6:04
  Speaker 1: I guess it doesn't, but he doesn't take a very
• 6:06
  Speaker 1: big sip.
• 6:07
  Speaker 4: I don't know. Yeah.
• 6:09
  Speaker 2: Another thing I think is later in the video, you
• 6:11
  Speaker 2: can see a screen on the barista's tablet. People were
• 6:16
  Speaker 2: zooming in and saying that the year said 2024, but
• 6:20
  Speaker 2: that was actually a digital manipulation that was spreading online.
• 6:23
  Speaker 2: Someone edited it to appear like it was 2024.
• 6:27
  Speaker 1: One right-wing influencer posted another theory on X about why
• 6:30
  Speaker 1: the video is fake.
• 6:31
  Speaker 3: His name is Matt Wallace.
• 6:33
  Speaker 2: He is very popular among the maga space on X
• 6:37
  Speaker 2: and also the conspiracy theory realm.
• 6:40
  Speaker 1: This is a post on X. At first, I thought
• 6:43
  Speaker 1: the Benjamin Netanyahu rumors were kind of silly. Sounds crazy, right?
• 6:46
  Speaker 1: But after watching this part of his coffee video, I
• 6:48
  Speaker 1: am genuinely starting to question if they are using a
• 6:51
  Speaker 1: highly advanced AI program. I slowed down this part where
• 6:54
  Speaker 1: it almost looks like his coat pocket glitches. So this
• 6:58
  Speaker 1: is I think he this is what he's talking about
• 6:59
  Speaker 1: the pocket here. I don't even know. Do you know
• 7:07
  Speaker 1: what to make of this? He's like slowed down, like
• 7:10
  Speaker 1: Nanyang is putting his left hand in his pocket and
• 7:12
  Speaker 1: this poster is claiming there's a glitch. I don't really
• 7:14
  Speaker 1: understand though.
• 7:15
  Speaker 2: I'm not 100% sure what he's referring to in terms
• 7:18
  Speaker 2: of a glitch. I will say that we do tend
• 7:20
  Speaker 2: to see a lot of this, especially from more conspiratorial
• 7:24
  Speaker 2: oriented social media users.
• 7:27
  Speaker 1: A lot of this, what does that mean? Like claims
• 7:29
  Speaker 1: of pocket glitches or what do you mean?
• 7:30
  Speaker 2: Not pocket glitches exactly, but very similar. They will take
• 7:34
  Speaker 2: a screenshot from a photo or they'll slow down a
• 7:38
  Speaker 2: clip from a video. And instead of like making a
• 7:40
  Speaker 2: definitive claim about what they're showing, they'll just say, take
• 7:43
  Speaker 2: a good look at this. Like this, does this open
• 7:46
  Speaker 2: your eyes? Like very generic claims that are impossible for
• 7:50
  Speaker 2: us to fact check. So, you know, we see a
• 7:52
  Speaker 2: lot of this type of conspiratorial twisting of facts. This
• 7:57
  Speaker 2: really helps these accounts get engagement. I'm sure if you
• 8:01
  Speaker 2: look at the comments, a lot of people may be asking, like,
• 8:04
  Speaker 2: what are you trying to show here?
• 8:05
  Speaker 3: I'm confused. You know, it creates a lot of engagement.
• 8:09
  Speaker 3: So that could be what's going on here.
• 8:12
  Speaker 1: Sophia says she and her team spent about a day
• 8:14
  Speaker 1: reporting out the veracity of this video of Netanyahu in
• 8:17
  Speaker 1: the cafe. They found several things that prove the video
• 8:20
  Speaker 1: is real.
• 8:20
  Speaker 2: For one, after this was posted by Netanyahu, Reuters did
• 8:24
  Speaker 2: a detailed analysis where they looked at their stock footage
• 8:28
  Speaker 2: from this same Jerusalem cafe and found that the background
• 8:32
  Speaker 2: of the video that Netanyahu posted matched their stock images
• 8:37
  Speaker 2: of this cafe. But more than that, the cafe itself
• 8:41
  Speaker 2: posted many different photos and videos of Netanyahu's interaction at
• 8:45
  Speaker 2: the cafe on their Instagram story and on their main
• 8:48
  Speaker 2: feed that were able to prove to us that this
• 8:52
  Speaker 2: actually was an authentic interaction.
• 8:55
  Speaker 1: Yeah, so here's their Facebook post from the day he
• 8:57
  Speaker 1: posted this. March 15th, 1024 in the morning, Jerusalem, Israel,
• 9:02
  Speaker 1: the Setaf Cafe. We were very happy to host the
• 9:04
  Speaker 1: prime minister and his office and staff today. And then
• 9:08
  Speaker 1: here are shots of him, like, sure enough, drinking this coffee,
• 9:11
  Speaker 1: same outfit, right? He's buying the coffee here. He's outside,
• 9:16
  Speaker 1: basking in the sun with the coffee.
• 9:19
  Speaker 2: Yeah, so there would have to be a lot of
• 9:20
  Speaker 2: people in on this in order for this to be
• 9:23
  Speaker 2: an AI-generated video.
• 9:25
  Speaker 3: This is actually really interesting, what you're showing on the
• 9:27
  Speaker 3: screen here.
• 9:27
  Speaker 1: Yeah, I'm just going to then grokwaden. At one point,
• 9:31
  Speaker 1: as I was scrolling through a thread on X about
• 9:33
  Speaker 1: this not-fake Netanyahu video, I saw that someone had asked
• 9:36
  Speaker 1: X's AI program, grok, if the video was real. And
• 9:40
  Speaker 1: grok had responded, I'm reading it here. This video is
• 9:44
  Speaker 1: AI generated. The casual cafe setting, lighting inconsistencies on the
• 9:48
  Speaker 1: face and hands, and the quote unquote still alive framing
• 9:50
  Speaker 1: don't match any verified Netanyahu appearances. It fits the wave
• 9:54
  Speaker 1: of deep fakes amid death hiding rumors in the Iran conflict.
• 9:57
  Speaker 1: No official confirmation or location match.
• 9:59
  Speaker 2: Yeah, so there's several false claims just in that post alone.
• 10:03
  Speaker 2: To say that there's no location match, that is just
• 10:07
  Speaker 2: not true. The location very clearly matches. So Grok is
• 10:11
  Speaker 2: integrated into X and anyone can respond to any post
• 10:15
  Speaker 2: with tagging the Grok X profile and asking like, at Grok,
• 10:21
  Speaker 2: is this real?
• 10:22
  Speaker 3: Is this AI generated?
• 10:24
  Speaker 2: At NewsGuard, we've done a lot of reports on this
• 10:27
  Speaker 2: feature that is integrated into X and showed how Grok,
• 10:31
  Speaker 2: the AI account, is actually one of the biggest spreaders
• 10:35
  Speaker 2: of false claims on this platform. You know, XAI, X
• 10:39
  Speaker 2: the platform, does not purport that their model is able
• 10:43
  Speaker 2: to accurately fact check false claims or that it's able
• 10:46
  Speaker 2: to accurately detect when something is AI-generated or not. But
• 10:51
  Speaker 2: people are consistently asking this chatbot to weigh in on
• 10:55
  Speaker 2: these issues, and it provides definitive answers despite not having
• 11:00
  Speaker 2: the tools in order to make these assessments. So, you know,
• 11:04
  Speaker 2: Grok is relying on what other social media users on
• 11:08
  Speaker 2: X are saying about a given topic, and it can
• 11:11
  Speaker 2: very easily be fooled into spreading false claims like this.
• 11:15
  Speaker 1: But with this lie about the Netanyahu video... It wasn't
• 11:18
  Speaker 1: only Grok that helped give it credence. A generally trusted
• 11:22
  Speaker 1: AI detection company, not Grok, but a service that is
• 11:25
  Speaker 1: specifically built for organizations like NewsGuard to use to detect deepfakes.
• 11:30
  Speaker 1: It got it wrong too.
• 11:31
  Speaker 2: There's an AI detector company named Hive. You could put
• 11:36
  Speaker 2: a video through their software and they run one test
• 11:40
  Speaker 2: on it in order to determine whether or not it's
• 11:42
  Speaker 2: likely to be AI generated.
• 11:44
  Speaker 3: So when users were.
• 11:45
  Speaker 2: Uploading this video into Hive, it was coming up as
• 11:49
  Speaker 2: over 95% likely that it was AI generated. So that,
• 11:54
  Speaker 2: you know, was being cited as a pretty clear definitive
• 11:57
  Speaker 2: proof for some people that this video was AI.
• 12:01
  Speaker 1: We're looking at like a tweet here from somebody saying,
• 12:04
  Speaker 1: an app for detecting artificial intelligence showed that the video
• 12:07
  Speaker 1: of Benjamin Netanyahu was 96.9% created with AI and is
• 12:09
  Speaker 1: not real. So the question arises, where exactly did Netanyahu
• 12:15
  Speaker 1: drink this coffee? And then it has a screenshot of
• 12:18
  Speaker 1: something from this place called Hive, I guess?
• 12:21
  Speaker 3: Exactly.
• 12:21
  Speaker 2: So this is a tool that they advertise as being
• 12:24
  Speaker 2: able to detect AI, but the company admits this is
• 12:27
  Speaker 2: not definitive proof that something is or isn't AI.
• 12:31
  Speaker 3: These models are not infallible.
• 12:34
  Speaker 2: I will say that at NewsGuard, Hive is one of
• 12:37
  Speaker 2: the tools that we sometimes depend on in order to
• 12:40
  Speaker 2: make an initial assessment on whether a video or a
• 12:43
  Speaker 2: photo is AI.
• 12:44
  Speaker 3: Generated or not.
• 12:46
  Speaker 2: Sometimes it's accurate and it points us in a good direction,
• 12:49
  Speaker 2: but we never rely on one software or an AI
• 12:53
  Speaker 2: detector as our only indicator of whether a piece of
• 12:57
  Speaker 2: content is or isn't AI generated because there's a lot
• 13:00
  Speaker 2: of ways these models can be tripped up.
• 13:02
  Speaker 1: I mean, to me, it kind of validates that this
• 13:05
  Speaker 1: video could be seen as AI generated, could be seen
• 13:08
  Speaker 1: as a deepfake. Like when I watched it kind of
• 13:11
  Speaker 1: quickly at first, I was like, yeah, this could be.
• 13:13
  Speaker 1: I think partly what does it for me is he's
• 13:14
  Speaker 1: very well lit. He's almost glowy, kind of. And that
• 13:19
  Speaker 1: gives it to me the sense of AI.
• 13:21
  Speaker 2: The video definitely looks like to me personally, like it
• 13:24
  Speaker 2: has some sort of filter on it.
• 13:26
  Speaker 3: So this is complete speculation.
• 13:28
  Speaker 2: But if I had to make an assumption of what's
• 13:31
  Speaker 2: going on, There are some cameras, like I know iPhone
• 13:34
  Speaker 2: has a feature called portrait mode, where it slightly blurs
• 13:38
  Speaker 2: out the background and highlights the individual who's the.
• 13:41
  Speaker 3: Main focus point of the video.
• 13:44
  Speaker 2: It appears to me that there could be some sort
• 13:46
  Speaker 2: of filter on it like that that could have tripped
• 13:49
  Speaker 2: up these AI detection models. But at NewsGuard, we tested
• 13:53
  Speaker 2: it on several.
• 13:54
  Speaker 3: There is a company called GetRealSecurity.com.
• 13:58
  Speaker 2: And they ran it through their models and their software,
• 14:01
  Speaker 2: and they determined that it was not AI generated. And
• 14:05
  Speaker 2: for everything that we verify, not just this video, we're
• 14:08
  Speaker 2: relying on more than just detection models because we know
• 14:11
  Speaker 2: that they're not infallible. I think that this situation really
• 14:14
  Speaker 2: opened up.
• 14:15
  Speaker 3: Our eyes to this problem.
• 14:18
  Speaker 2: Just six months ago, when we were using Hive and
• 14:20
  Speaker 2: other detection models, we weren't really seeing a lot of
• 14:23
  Speaker 2: these false positives or false negatives. It was usually pretty
• 14:27
  Speaker 2: good at assessing whether something was AI or not.
• 14:30
  Speaker 3: And I think that AI.
• 14:31
  Speaker 2: Content has gotten so much more realistic in just such
• 14:35
  Speaker 2: a short amount of time. Just in October or September,
• 14:39
  Speaker 2: OpenAI released its Sora 2 video generator. And that massively
• 14:45
  Speaker 2: changed the accessibility of people to be able to produce
• 14:49
  Speaker 2: these highly realistic deepfakes in a matter of seconds with
• 14:53
  Speaker 2: just a very simple text prompt. We've seen these tools
• 14:57
  Speaker 2: attempt to adapt, but maybe in this case, they're overcorrecting.
• 15:02
  Speaker 1: Yeah, I mean, pardon my language, but it's a fucking
• 15:04
  Speaker 1: mess is what it sounds like. It's a mess.
• 15:06
  Speaker 3: It's such a mess.
• 15:07
  Speaker 2: The fact that a well-known detection tool like Hive said
• 15:12
  Speaker 2: it was AI, if you're just a normal person and
• 15:15
  Speaker 2: you see a detection tool, a company that purports that
• 15:19
  Speaker 2: this is what they do, they are able to detect
• 15:21
  Speaker 2: AI and they're calling it AI generated, it's hard to
• 15:25
  Speaker 2: argue with that.
• 15:26
  Speaker 1: So interesting with this false claim of the week is
• 15:28
  Speaker 1: because I get the sense that a lot of times
• 15:31
  Speaker 1: what you guys are correcting is deep fakes. Like there
• 15:34
  Speaker 1: are deep fakes that are faking something and you're having
• 15:36
  Speaker 1: to jump in and say, this is fake. This didn't
• 15:38
  Speaker 1: really happen. But in this case, it's the inverse. It's
• 15:43
  Speaker 1: a real video that people are claiming is a deep fake.
• 15:46
  Speaker 1: And it's just as insidious. Like it's just as bad.
• 15:48
  Speaker 3: You're totally on point there.
• 15:50
  Speaker 2: I would say that, you know, it started off with us,
• 15:54
  Speaker 2: especially in relation to this war in Iran. We were
• 15:58
  Speaker 2: fact-checking so much AI-generated content. It was unlike something I've
• 16:03
  Speaker 2: seen during my time at this company.
• 16:05
  Speaker 4: Really?
• 16:06
  Speaker 2: Just the sheer volume of fake videos and photos that
• 16:09
  Speaker 2: were being spread online.
• 16:10
  Speaker 3: A lot of it were purporting to show missile.
• 16:14
  Speaker 2: Strikes on major cities in the Middle East, massive destruction.
• 16:18
  Speaker 3: It's a very confusing time.
• 16:20
  Speaker 2: A lot of people are turning to social media to
• 16:24
  Speaker 2: really grasp what's going on And I think that a
• 16:27
  Speaker 2: lot of bad actors understand that and they are able
• 16:32
  Speaker 2: to produce these videos that can get massive traction on
• 16:36
  Speaker 2: platforms like X, Instagram, you know, all of the major
• 16:39
  Speaker 2: platforms and really manipulate public opinion about what's going on.
• 16:43
  Speaker 2: So I think that this is like the perfect storm
• 16:46
  Speaker 2: for this conflict right now. Wow.
• 16:48
  Speaker 1: So you feel like the last three weeks, this conflict
• 16:51
  Speaker 1: is a new theater of war, basically, like you're on
• 16:54
  Speaker 1: the front lines of something new right now. We all are.
• 16:57
  Speaker 2: I definitely think so. For example, I'm thinking of this
• 17:00
  Speaker 2: one video that we actually named as our false claim
• 17:03
  Speaker 2: of the week two weeks ago. It was an AI
• 17:06
  Speaker 2: generated video, but it was being presented.
• 17:08
  Speaker 3: As if it was authentic.
• 17:10
  Speaker 2: And it showed dozens of Iranian missiles striking a center,
• 17:15
  Speaker 2: like a major hub in Tel Aviv and causing massive destruction. Now,
• 17:21
  Speaker 2: at that point in the war, this was like the
• 17:24
  Speaker 2: first couple of days, There was only one missile that
• 17:27
  Speaker 2: was actually able to go past the Israeli air defense
• 17:30
  Speaker 2: system and strike Tel Aviv, and it didn't cause massive destruction.
• 17:35
  Speaker 2: But this video went mega viral.
• 17:38
  Speaker 3: We saw multiple posts on.
• 17:40
  Speaker 2: X that had over like 50 million views, and it
• 17:43
  Speaker 2: looked believable. There was very few signs in that video
• 17:47
  Speaker 2: that you can definitively say is not real. Social media
• 17:51
  Speaker 2: users are able to twist that. We saw people commenting
• 17:54
  Speaker 2: on this video and saying, oh, the mainstream media is
• 17:58
  Speaker 2: hiding from us that Iran had such a major success
• 18:01
  Speaker 2: in Tel Aviv. But when you look at on the
• 18:04
  Speaker 2: ground reporting in Tel Aviv, we know that a missile
• 18:07
  Speaker 2: strike of that scale didn't occur. So these, you know,
• 18:10
  Speaker 2: AI videos are able to further undermine trust in our
• 18:15
  Speaker 2: media that's actually on the ground in this region reporting
• 18:18
  Speaker 2: on the conflict. But at the same time as that,
• 18:21
  Speaker 2: we're also seeing, obviously, what we've just been talking about
• 18:24
  Speaker 2: is that it allows people to discount authentic footage of
• 18:28
  Speaker 2: the war and say anything that goes against their worldview
• 18:32
  Speaker 2: is AI.
• 18:33
  Speaker 1: I interviewed someone not long ago who's an expert in
• 18:36
  Speaker 1: social media's use in warfare. And he was talking about
• 18:39
  Speaker 1: how he got involved in that work with the conflict
• 18:42
  Speaker 1: between Israel and Hamas back in 2012 and how that
• 18:45
  Speaker 1: was referred to as the first Twitter war. basically, because
• 18:48
  Speaker 1: it was the first time they saw things that were
• 18:50
  Speaker 1: happening on Twitter actually driving decisions made by the IDF
• 18:55
  Speaker 1: and such. And hearing you talk, I wonder if what
• 18:58
  Speaker 1: we're talking about is the first AI war right now.
• 19:01
  Speaker 2: I'm definitely not a war expert, but from just what
• 19:04
  Speaker 2: I've observed at NewsGuard during my time here, the amount
• 19:08
  Speaker 2: of AI content being used to manipulate public perception of
• 19:13
  Speaker 2: what's going on in the war is It definitely seems
• 19:15
  Speaker 2: like a new frontier that we haven't crossed previously. It
• 19:19
  Speaker 2: is really crazy because something we've been telling our readers
• 19:24
  Speaker 2: and writing about is the fact that when you're on
• 19:26
  Speaker 2: social media, you can't really believe your eyes anymore because
• 19:31
  Speaker 2: AI content is just getting so realistic. And a lot
• 19:35
  Speaker 2: of times we'll find that social media users don't take
• 19:38
  Speaker 2: that advice and they'll readily believe that AI content is real.
• 19:43
  Speaker 2: But we're kind of seeing, again, the reverse effect happening
• 19:47
  Speaker 2: here where you can you know, kind of say that
• 19:50
  Speaker 2: anything is AI nowadays and it could be somewhat believable.
• 19:53
  Speaker 1: Yeah, it's like the AI has gotten so good at
• 19:58
  Speaker 1: being real that the real stuff now looks fake.
• 20:03
  Speaker 3: Yeah, no, it's kind of messed up.
• 20:05
  Speaker 1: Of all the many Deep fakes and things that people
• 20:09
  Speaker 1: were claiming were deep fakes that weren't and lies and
• 20:12
  Speaker 1: bullshit and stuff that you waded through this week. Why
• 20:15
  Speaker 1: was this one, the video of Benjamin Netanyahu, the false
• 20:18
  Speaker 1: claim of the week? Why did this get the title?
• 20:21
  Speaker 2: It was the sheer amount of engagement of this claim.
• 20:25
  Speaker 2: We have a tool that we use in order to
• 20:28
  Speaker 2: assess verified engagement metrics on social media. And we found
• 20:33
  Speaker 2: that it had over 50 million views. verified views and
• 20:37
  Speaker 2: the actual number is much, much higher than that, that's
• 20:40
  Speaker 2: actually somewhat unusual for us to have that high of
• 20:44
  Speaker 2: a view count.
• 20:46
  Speaker 1: After Sophia and I talked, this was on Friday, I
• 20:49
  Speaker 1: saw another engagement monger give this fake story even more
• 20:52
  Speaker 1: of a boost. Joe Rogan talked about it on his podcast.
• 20:55
  Speaker 4: What do you think about these Netanyahu AI videos?
• 20:59
  Speaker 1: I haven't seen them. You haven't seen them?
• 21:01
  Speaker 4: No. They think he might be dead. What? Yeah. There's
• 21:05
  Speaker 4: a bunch of AI videos that Israel has released that
• 21:08
  Speaker 4: are like clearly AI.
• 21:10
  Speaker 3: What?
• 21:11
  Speaker 4: Show him the one where there's in the cafe. This
• 21:15
  Speaker 4: one's nuts.
• 21:17
  Speaker 1: He went through the same questionable elements of the video.
• 21:19
  Speaker 4: First of all, it's weird because he sips out of
• 21:22
  Speaker 4: the cup and yet the cup stays exactly the same level.
• 21:27
  Speaker 4: And no matter where he moves the cup around, it
• 21:30
  Speaker 4: doesn't spill.
• 21:32
  Speaker 1: I admit. In the early hours of this claim about
• 21:35
  Speaker 1: the Netanyahu video being fake, especially with an AI detection
• 21:38
  Speaker 1: tool saying it was probably fake, I could see getting
• 21:42
  Speaker 1: drawn into it. But by the time Joe Rogan ran
• 21:45
  Speaker 1: with this, Sophia and her team had put out their
• 21:47
  Speaker 1: fact check. A bunch of other places had too. Reuters,
• 21:50
  Speaker 1: Times of Israel, Snopes. The photos from the cafe showing
• 21:54
  Speaker 1: Netanyahu had been there were publicly available and reported on.
• 21:58
  Speaker 1: A reporter for an Indian TV station had visited the
• 22:00
  Speaker 1: cafe to confirm that the prime minister had been there.
• 22:03
  Speaker 1: But none of that stopped Rogan from sharing his conclusion
• 22:06
  Speaker 1: about Netanyahu with his 20 million plus listeners.
• 22:09
  Speaker 4: He might be dead.
• 22:11
  Speaker 1: I tried messaging Rogan to tell him about the fact checking,
• 22:14
  Speaker 1: to ask if he'd seen it or not, and to
• 22:17
  Speaker 1: see if he'll update his audience. He didn't get back
• 22:19
  Speaker 1: to me. So you guys looked at the metrics for
• 22:23
  Speaker 1: this false claim and saw that there were at least
• 22:26
  Speaker 1: 50 million views of it across social media in the
• 22:30
  Speaker 1: last week, basically, probably much more. And then you guys
• 22:35
  Speaker 1: got to it and said about debunking it. And then
• 22:38
  Speaker 1: you put it into your database and your newsletter, which
• 22:41
  Speaker 1: has thousands, tens of thousands of readers. What?
• 22:44
  Speaker 3: Yeah, tens of thousands of readers.
• 22:46
  Speaker 1: So what are we doing here?
• 22:48
  Speaker 3: Yeah, no, I mean, that's a valid question. And it
• 22:51
  Speaker 3: can be discouraging to think of it that way. But
• 22:53
  Speaker 3: I also do think.
• 22:54
  Speaker 2: That it's important work to have people and journalists who
• 22:58
  Speaker 2: are working to verify this content. we're seeing a lot
• 23:02
  Speaker 2: of people just on X alone turning to Grok and saying,
• 23:07
  Speaker 2: is this AI?
• 23:08
  Speaker 3: Is this real?
• 23:09
  Speaker 2: And the only way that these tools like Grok and
• 23:12
  Speaker 2: other AI platforms and search engines can be accurate is
• 23:17
  Speaker 2: if there is accurate information out there. A lot of
• 23:20
  Speaker 2: times we'll see that Grok is responding incorrectly to users
• 23:26
  Speaker 2: asking about a claim.
• 23:28
  Speaker 3: And then once.
• 23:29
  Speaker 2: NewsGuard hosts our fact check of that claim online, Grok
• 23:33
  Speaker 2: will then begin to reference NewsGuard and say that according
• 23:36
  Speaker 2: to NewsGuard, this claim isn't true. So that's what motivates me.
• 23:40
  Speaker 3: I think it's still important work.
• 24:00
  Speaker 1: Sophia Rubinson edits the Reality Check newsletter from NewsGuard. After
• 24:04
  Speaker 1: this interview, out of curiosity, I asked Grok if the
• 24:08
  Speaker 1: video of Netanyahu was authentic. And sure enough, at this point,
• 24:11
  Speaker 1: days after the fact, Grok told me it was authentic.
• 24:14
  Speaker 1: We tested it a few times with different members of
• 24:16
  Speaker 1: our team. Same result. I'm happy to report that Grok
• 24:20
  Speaker 1: is now correct when it comes to this video of
• 24:22
  Speaker 1: Netanyahu drinking coffee. Though I will say, at one point,
• 24:26
  Speaker 1: Grok told us it did use NewsGuard's reporting to help
• 24:29
  Speaker 1: make that determination. And other times it said it did
• 24:32
  Speaker 1: not use NewsGuard. So, it's still hard to know what's true.
• 24:36
  Speaker 1: And something else happened this week after I spoke to Sophia.
• 24:39
  Speaker 1: OpenAI shut down its AI video generation app, Sora, like overnight.
• 24:44
  Speaker 1: It's kind of surprising. This is the app Sophia said
• 24:47
  Speaker 1: has been flooding the internet with so many realistic deepfakes
• 24:50
  Speaker 1: about the war. There are still other tools for making
• 24:53
  Speaker 1: AI videos, but it'll be interesting to see if things
• 24:55
  Speaker 1: change at all now that Sora's gone. Please, if you're
• 25:01
  Speaker 1: new to our show or if you haven't done this yet,
• 25:04
  Speaker 1: follow Question Everything on Apple or Spotify or wherever you
• 25:06
  Speaker 1: get your podcasts and rate and review us. It really
• 25:09
  Speaker 1: helps us get seen. Our newsletter is questioneverything.substack.com. You can
• 25:13
  Speaker 1: subscribe there. And you can reach me with tips, story ideas,
• 25:16
  Speaker 1: angry rants, jokes, at brihreid, B-R-I-H-reid on Instagram or on
• 25:22
  Speaker 1: Signal at brihreid. 45. Today's show was produced by Sophie Casis
• 25:27
  Speaker 1: and edited by our managing editor, Kevin Sullivan. Thanks to
• 25:30
  Speaker 1: the team at NewsGuard for working on this with us
• 25:32
  Speaker 1: and to Amir Yanai for translation help. Robin Semien and
• 25:35
  Speaker 1: I are the executive producers of Question Everything. Our team
• 25:38
  Speaker 1: also includes producer Zach St. Louis, contributing producers Sam Egan,
• 25:41
  Speaker 1: contributing editors Neil Drumming and Jen Kinney, and associate producer
• 25:45
  Speaker 1: Kevin Shepard. This episode was fact-checked by Annika Robbins, mixing
• 25:49
  Speaker 1: and sound design by Brendan Baker. Matt McGinley composed our music.
• 25:53
  Speaker 1: If you're interested in supporting Question Everything as a partner
• 25:55
  Speaker 1: or a sponsor, please write us at hey, H-E-Y, at placementtheory.com.
• 26:00
  Speaker 1: Our partners at KCRW include Arnie Seipel, Tejal Ajmera, Natalie Hill,
• 26:04
  Speaker 1: and Jennifer Farrow. Thanks for listening, and we will see
• 26:06
  Speaker 1: you next week.