The most powerful AI model ever just went public 🤯

Good morning. Anthropic just released Claude Fable 5, the most powerful AI model ever made available to the public. YouTube is now automatically detecting and labeling AI-generated videos even if creators don't disclose it. And a fake AGT contestant named Ernesto fooled over 50 million viewers before anyone realized nothing in the video was real. Have you tried Fable 5 yet? Hit reply and tell me. We cover all three below.

Anthropic released Claude Fable 5 on June 9, the first publicly available version of its Mythos model. This is the same model that was restricted to a handful of partners like AWS, Microsoft, and Apple since April because of how capable it is. Now anyone can use it.

Fable 5 is the best coding model in the world right now and its vision capabilities are a significant leap forward. It can analyze, understand, and reason about visual content at a level nothing else matches. For AI video creators, that matters because of MCP. Any tool that supports Model Context Protocol can now connect to Fable 5, meaning the most powerful AI model available is now a layer inside your creative workflow. You can use it to analyze footage, write complex prompts, build automation pipelines, or direct AI video generation from inside whatever app you already work in.

Fable 5 is available on Pro, Max, Team, and Enterprise plans through June 22 at no extra cost. After that, usage credits may be required. If you've been building any kind of AI video workflow that involves coding, automation, or visual reasoning, this is the model to test right now.

Starting this month, YouTube is using internal detection signals to automatically identify videos with significant photorealistic AI content. If you don't disclose it yourself, YouTube will apply the label for you. This is a major shift from the old system, which relied entirely on creators to self-report.

Labels now sit directly below the player on long-form videos and as an overlay on Shorts. They're permanent for anything made with YouTube's own AI tools like Veo or Dream Screen, and for any content with C2PA metadata indicating fully generative AI. Creators can contest labels they think are wrong through YouTube Studio, but the defaults are sticky.

Here's what matters: AI-labeled videos will not be penalized in the algorithm and will not lose monetization. The labels are informational, not punitive. YouTube also expanded its deepfake protections to all adults 18 and over, so anyone can now request removal of AI-generated content that uses their likeness. For creators making AI video content for YouTube, the move is simple. Disclose proactively in YouTube Studio before the system flags you. It looks better and you keep control of the narrative.

A video of a 54-year-old man named Ernesto performing a heartbreaking song called "Still Waiting at the Door" on America's Got Talent spread across TikTok, YouTube, and Instagram. Viewers watched a homeless carpenter sing about losing his family, judges wiping tears, and a standing ovation from the crowd. It racked up over 24 million views on YouTube and 27 million on TikTok. People were genuinely moved.

None of it was real. Ernesto doesn't exist. He never appeared on AGT. The contestant, the judges' reactions, the backstory, and the performance were all generated or manipulated using AI by a YouTube channel called AGTverseai. They combined real AGT footage with AI-generated graphics and voiceover. The channel ran the same playbook with other fake contestants and racked up millions more views each time.

For creators, this is the Ernesto paradox. The same tools that let you build an audience with AI-generated content can also build an audience built entirely on deception. AGT's producers had to publicly confirm Ernesto never existed. The video still has comments from people who believe it's real. AI video has crossed the line where emotional manipulation at scale is now a one-person operation.