Podcast Summary: A Video So Real It Looks Fake

Podcast: Question Everything
Host: Brian Reed
Guest: Sophia Rubinson, Senior Editor at NewsGuard
Date: March 26, 2026

Episode Overview

In this episode, Brian Reed delves into the complex and evolving landscape of online misinformation, focusing on a single viral lie: the claim that a recent “proof-of-life” video of Israeli Prime Minister Benjamin Netanyahu, posted amidst the war in Iran, was generated by AI. Reed and his guest, Sophia Rubinson of NewsGuard, explore how such claims take root, the challenges of debunking them, and why the boundaries between truth and fabrication are now more blurred than ever.

Key Discussion Points and Insights

The False Claim of the Week: Netanyahu’s “AI Video”

• Central Story: Iranian state media and pro-Iranian social media users claimed that a video of Netanyahu at a Jerusalem café was a deepfake, despite the fact that it was an authentic, unscripted video posted by Netanyahu himself.
• Background: (04:46) Rubinson describes how ongoing conflicts have flooded the internet with fake images and videos, creating an unprecedented challenge for verification.
• Layered Disinformation: (05:45) Reed and Rubinson map out the timeline and layers: rumors of Netanyahu’s death → Netanyahu posts proof-of-life video → new rumors claim that video is an AI fake.

Anatomy of the Viral Lie

• Breakdown of the Video: (06:17)
  • Netanyahu’s video: “They say I’m what? Watch.”—he sips coffee, shows his hand (in response to earlier, unrelated rumors about a supposed “sixth finger”), and cracks a joke.
  • Reed details the visual evidence, including other real-world photos and Instagram posts from the café itself confirming Netanyahu’s visit.
• Conspiratorial Analysis: (08:42)
  • Influencers like Matt Wallace amplify suspicions by pointing to supposed "glitches" in the video (e.g., the coffee level not changing or fabric oddities), using vague language that leverages uncertainty and boosts online engagement.
  • Quote: “They will take a screenshot... and instead of making a definitive claim... just say, take a good look at this—does this open your eyes?” (Rubinson, 09:41)

The Role of AI Detection Tools—And Their Failures

• AI Chatbots as Misinformation Vectors: (11:39)
  • X’s chatbot, Grok, initially declared the video AI-generated, referencing supposedly mismatched settings.
  • Quote: “Grok... is actually one of the biggest spreaders of false claims on this platform.” (Rubinson, 12:11)
  • NewsGuard has documented Grok’s inaccuracies and its tendency to echo prevailing social narratives regardless of their truth.
• Automated Detectors Also Get It Wrong: (13:26)
  • Hive, a leading AI-content detector, scored the Netanyahu café video as 96.9% likely to be AI.
  • Rubinson explains that such tools are “not infallible,” that NewsGuard uses them only in combination with other evidence, and that AI-generated content has recently outpaced detection technology.
  • Quote: “Just six months ago... we weren’t really seeing a lot of these false positives or false negatives. Now, AI content has gotten so much more realistic.” (Rubinson, 15:32)

The New Information Battlefield

• Conflict as a Deepfake Laboratory: (18:00)
  • The war between America, Israel, and Iran is called “a new theater of war”—not just militarily, but in information.
  • Volume and reach of fakes are at an all-time high.
  • Quote: “A lot of bad actors... are able to produce these videos that can get massive traction... it really manipulates public opinion.” (Rubinson, 18:00)
• Reverse Misinformation: (17:37)
  • Increasingly, real footage is dismissed as fake—sowing confusion and skepticism everywhere.
  • “In this case, it’s the inverse. It’s a real video that people are claiming is a deepfake. And it’s just as insidious... just as bad.” (Reed, 17:37–18:00)
• The “First AI War”?: (20:44)
  • As AI fakes become indistinguishable from reality, both authentic and artificial media are equally subject to doubt.

Why the Claim Mattered

• Massive Engagement: (22:32)
  • This false claim got “over 50 million verified views” in a week, making it one of NewsGuard’s most widely spread bits of disinformation to date.
• Amplification by Influencers: (23:07–24:22)
  • Joe Rogan, the world’s biggest podcaster, discussed the false claim on air, further amplifying it to 20 million listeners, despite easily available debunks from reputable sources.
  • Neither Rogan nor his co-hosts seemed interested in updates from NewsGuard, Reuters, Snopes, or actual on-location verification.

The Challenges and Importance of Fact-Checking

• NewsGuard’s Method: (24:59)
  • NewsGuard uses a combination of manual verification, reverse image searches, and multiple AI tools, but always emphasizes the need for human judgment.
• Effect on AI Tools: (25:21)
  • Fact checks published by NewsGuard and others eventually led Grok to update its answer, referencing accurate reporting instead of social buzz.
  • Quote: “That’s what motivates me. I think it’s still important work.” (Rubinson, 25:01)

Notable Quotes & Memorable Moments

• “It was like seeing the whole information war around America and Israel’s assault on Iran in a grain of sand.” — Brian Reed (02:22)
• “You can’t really believe your eyes anymore because AI content is just getting so realistic... but we’re kind of seeing, again, the reverse effect... where you can kind of say that anything is AI nowadays and it could be somewhat believable.” — Sophia Rubinson (21:16)
• “It’s a fucking mess is what it sounds like. It’s a mess.” — Brian Reed (17:13)
• “The real stuff now looks fake.” — Brian Reed (22:05)
• “The only way that these tools... can be accurate is if there is accurate information out there.” — Sophia Rubinson (25:01)

Timestamps for Key Segments

• 03:12: Rubinson on unprecedented flood of AI fakes in current conflict
• 04:46–06:15: Breakdown of the Netanyahu café video and related conspiracy theories
• 08:42–09:41: Viral posts and how influencers drive engagement with ambiguity
• 11:39–12:11: Grok and the risk of using AI chatbots for fact-checking
• 13:26–15:32: Flaws in AI detection tools like Hive; why human oversight matters
• 18:00–18:59: Social media as a battlefield in the ongoing war
• 20:44–21:16: Questioning whether this is the “first AI war”
• 22:32: Why this particular claim became the “False Claim of the Week”
• 23:07–24:22: Rogan’s amplification and the persistence of the lie
• 24:59–26:11: The slow impact (and importance) of professional fact-checking

Overall Tone

Candid, occasionally exasperated, but persistent in seeking clarity and truth. The conversation is informed, skeptical, and grounded in both the technical realities and the broader moral questions of journalism today.

Takeaway

This episode weaves together the specifics of the Netanyahu café video story and the broader, more troubling reality: misinformation and AI-powered manipulation are transforming not just our interpretations of world events, but our very ability to trust what we see. Fact-checking is harder and more essential than ever, but even then, it’s a game of constant catch-up in an ecosystem where fakes and doubts alike go viral in an instant.

“The AI has gotten so good at being real that the real stuff now looks fake.” — Brian Reed (22:05)