AI Summary
5 min readApple released a folding iPhone with anti-deepfake features, OpenAI shipped a new image generator with a sketch-to-image tool, an Anthropic researcher quit with a warning that AI could kill everyone, Suno launched a licensed music model, hackers are draining Claude token allowances, and Meta debuted a personal AI agent with a Stripe integration and a $130,000 bug bounty. It was a dense day in AI.
Apple’s Anti-Fake Camera and Folding Phone
Apple announced a reference image system tied to the iPhone 18 Pro launch. When a photo is taken, signed sensor data is captured and routed through Apple’s private cloud compute, producing what the company says is an unalterable image. Apple is adopting the C2PA content credentials standard to help users detect AI-generated or AI-altered images. The iPhone 18 Pro will have an 84-megapixel fusion main camera with a variable aperture built from six laser-cut blades. A reference image API is coming for third-party developers. The new folding iPhone Duo has screens on both sides, which led to a viral meme about how the rear-facing selfie preview will let subjects critique the photographer’s framing — "boyfriends are cooked," as one comment put it.
OpenAI’s Image 2.5 and Sketch
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Hard Fork AI
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **Episode Kickoff & Host's Opening Remarks** - The host introduces a busy day in tech and AI news, teasing a packed agenda of stories.
- 2 (00:11) **Apple's New iPhone Duo & Reference Image Feature** - Apple announces a folding iPhone with AI features aimed at combating deepfakes.
- 3 (01:01) **OpenAI Launches Image 2.5 with "Sketch" Feature** - A new image generation model is released, featuring a powerful "Sketch" tool for turning doodles into images.
- 4 (01:17) **Anthropic Researcher Jacob Coxton Quits, Warns of AI Existential Risk** - A former researcher goes viral with a post claiming AI could kill all humans by 2030.
- 5 (01:29) **Suno Launches V6 Model with Licensed Music** - After facing multiple copyright lawsuits, Suno releases a new model trained solely on licensed data.
- 6 (01:30) **Hackers Draining Claude Subscriber Token Allowances** - Stolen sessions and malware are being used to hijack user accounts and consume their paid tokens.
- 7 (01:53) **Meta Launches "Muse" Personal AI Agent with Security-First Pitch** - A new AI agent is released with a massive bug bounty and direct payment integration.
+ Full timestamped outline available in the app
Guests on this episode
Show Notes
In this episode, we cover insights shared by an Anthropic researcher on AI's potential dangers. We also explore the innovations from OpenAI's new image model.
More from this podcast
Hard Fork AI →