AI Made This Entire Video by Itself... (Claude Fable 5)

Summary
Content creator Dan Dingle tests Anthropic's Claude Fable 5 by prompting the model to generate synthetic video clips using "Seedance 2.0," create an AI clone of his face and voice to react to them, and automatically edit the final video in his signature style. The real Dan Dingle watches and comments on the AI-generated video, critiquing the oddities, hallucinations, and pacing of his digital double.

What is shown

• [00:03] A BBC News article headline: "Anthropic suspends new AI tools over US government security concerns" (dated 13 June 2026).
• [00:17] Prompt interface ("Evening") showing the prompt: "Generate AI videos then have an AI version of myself react to them in an 'entertaining' YouTube video. MAKE NO MISTAKES." with a system tag noting: "Fable 5 is the most capable model and draws down usage much faster than Opus 4.8".
• [01:11] AI video segment featuring a gym bro bench-pressing a barbell that morphs into spaghetti while the spotter gives a thumbs-up.
• [02:17] AI video segment: "CCTV of a horse doing a performance review over the phone", showing an anthropomorphic horse in an office cubicle typing with hooves and discussing "synergy."
• [03:41] AI video segment: A golden retriever driving a taxi through New York City with a passenger looking terrified and surreal AI voice hallucination mentioning "Jeremy."
• [04:44] AI video segment: A TV chef flips a pancake that vanishes into the sky; the presenter's avatar warns, "Remember this pancake. It matters later."
• [05:32] AI video segment: A bouncy castle floats into the sky and suburban fathers pursue it, with one harpooning it using a garden hose.
• [06:06] AI video segment: A news studio flooded with orange juice with an anchor remaining calm.
• [06:51] AI video segment: Police bodycam footage arresting a mime for a noise complaint while trapped inside a visible glass/invisible box.
• [07:29] "The Final Prompt" combining all previous scenes into one New York street sequence, culminating in the airborne pancake landing squarely on the mime's head ([08:19]).

Claims & numbers

• The presenter claims Claude Fable 5 is "the world's most powerful AI right now" and was "literally banned by the US government for a couple of weeks" before being restored ([00:01]).
• The interface banner states that "Fable 5 is the most capable model and draws down usage much faster than Opus 4.8" ([00:17]).
• The AI presenter claims "Seedance 2" recently dropped with native audio generation ([00:38]).

Notable quotes

• [00:41] AI Dan: "The slop has a voice. We have to look."
• [05:14] AI Dan: "Remember this pancake. It matters later."
• [08:24] AI Dan: "Two setups, one payoff. Cinema."

Assessment
This is an authentic entertainment/reaction video by a creator testing an agentic video generation pipeline. The embedded reaction video features an AI avatar and voice clone responding to AI-generated surreal clips with characteristic procedural generation artifacts, speech hallucinations, and jerky timing.

Lyrics & themes
The AI-generated video is narrated as a comedic reaction show divided into thematic rounds:

• Round 1 (Animals with Careers): Features gym bro spaghetti lifting, an office horse doing performance reviews ("More synergy moving forward..." at [02:38]), and a dog cab driver ("Jeremy, blink twice if the dog is talking" at [04:03]).
• Round 2 (Physics Crimes): Surreal physical violations including an ascending pancake, floating bouncy castle ("He hose-harpooned it" at [05:40]), and an orange juice news flood.
• The Final Prompt: Narrative payoff combining all previous prompt elements into a single multi-character climax.

Lore & references

• Seedance 2.0: A reference to ByteDance's generative video model, parodying native video-to-audio generation tools.
• The Spaghetti Bench Press: An homage to the classic AI video benchmark meme originating from early Will Smith eating spaghetti clips.
• "The Slop": Community slang for low-effort or bizarre generative AI video outputs.
• The Chekhov's Pancake: A parody of narrative foreshadowing, where the disappearing pancake from round 2 returns in the finale to land on the mime.

Visual style & craft

• Reaction overlay: The AI-generated presenter mimics Dan Dingle's home studio setup with purple/pink ambient lights, Carhartt shirt, and microphone, but exhibits telltale deepfake traits: stiff neck movements, repetitive hand gestures, and an unnerving frozen smile.
• Video clips: Highly polished yet surreal diffusion video artifacts typical of 2026 video models, including morphing geometry (spaghetti bar), hallucinated text in graphics and ticker bars, inconsistent timecode counters, and rapid, chaotic editing rhythms.

Described by gemini-3.8-flash on 2026-09-29 from the video's audio and frames.