Apple wants to prove your iPhone photos aren't AI slop

Share
Apple wants to prove your iPhone photos aren't AI slop

Apple's launch event was mostly hardware, but the most consequential announcement was defensive: a way to prove a photo came out of a camera and not a model. Here's what shipped, and what it changes.

Apple is shipping cryptographic proof of provenance on the iPhone 18 Pro. When you press the shutter, the main camera sensor also captures signed sensor data, and Apple's Private Cloud Compute turns it into a reference image — Apple's own phrase is a "digital negative" — that you can view in Photos next to the edited version, so any later change is detectable by comparison. Developers get APIs for it, and Apple says it will also support Google's SynthID standard, which is how AI-generated and AI-edited images get tagged in the first place. The pitch is aimed squarely at working photographers: Apple called the feature "vital for photojournalists and photographers," and it is not hard to see why — a newsroom that can hand an editor a signed frame has something a model cannot fake.

The catch is that provenance only works when the whole chain honors it, and Apple's version has to survive leaving the Photos app. Signed sensor data proves what the lens saw; it does not stop anyone from screenshotting the image, re-encoding it, or posting it somewhere that strips the reference. We have been here before — we showed last month that C2PA camera proofs can be forged on a rooted Pixel, and the same lesson applies: signing is easy, tamper-evidence across the open internet is not. Supporting SynthID alongside its own standard is the right call, because a phone maker's private format is worthless the moment a picture lands on a social network.


The Apple Watch is now taking notes on your conversations. Series 12 and Ultra 4 ship with "Audio Intelligence": Live Rewind shows the last 15 seconds of a conversation as text when you double-press the Digital Crown, and Siri Recap uses ambient listening to write high-level takeaways of a meeting or a parent-teacher conference into the new Siri app. Apple says the features create no audio recordings. Alongside them, the new health-sensing system samples heart rate every five seconds — 60 times more often than before — and adds readiness scores and a "Health Age" estimate. Series 12 starts at $399, Ultra 4 at $499.


The iPhone Duo is Apple's first foldable, and it is an on-device AI bet. The $1,999 device pairs a 5.4-inch outer screen with a 7.6-inch inner display and runs the A20 Pro, whose new dual 16-core Neural Engine Apple says delivers twice the compute of the previous generation for on-device models. iOS 27 gives it real Split View multitasking for the first time on iPhone, and Siri AI — the rebuilt assistant arriving in beta — uses onscreen awareness to act on whatever is on the larger display. Pre-orders open October 16, shipping October 23.

What to watch: whether any major social platform agrees to carry Apple's reference image, or whether this stays a feature that only works inside Photos.

Would you trust a signed sensor photo more than an unsigned one — or does provenance only matter once every platform honors it? Tell us in the comments.

Read more

OpenAI drops 372 math results, nearly all from one prompt

OpenAI drops 372 math results, nearly all from one prompt

A dense news day across three fronts: OpenAI published a mountain of mathematics from an unreleased model, SpaceX reportedly wants the debt market to fund its Nvidia orders, and Anthropic restructured its cyber program around three tiers of trust. OpenAI published 372 mathematical result families from an internal frontier model — and says nearly all of them came from a single prompt handed to a single AI agent. The release comprises 722 manuscripts touching hundreds of open questions, plus Lea

Underdog launches a private on-device AI assistant, backed by a16z

Underdog launches a private on-device AI assistant, backed by a16z

The privacy split in consumer AI got a new entrant tonight, GitHub turned code-review benchmarks into a vendor-bias argument, and OpenAI opened its oddest API to everyone. Underdog launched in invite-only beta: an on-device AI assistant that keeps your data on your machine and never charges you a subscription. Self-taught coder and Thiel Fellow Sigil Wen — who moved to Silicon Valley at 17 and lived in an AI hacker house with Andrej Karpathy — built his own inference engine, Husky, to run a 2

Today in AI — October 6, 2026

Today in AI — October 6, 2026

A day of second-order moves: the labs are buying task data from software vendors instead of scraping the web, Waymo is borrowing to fund the robot world, and the biggest bank in the US just put a number on what Anthropic's latest model cost it in risk. Models & Research * OpenAI is training GPT-6 Astra on Ironclad's real contracting work. Ironclad staff helped turn 11 tasks across legal, commercial and procurement work — setting up NDAs, approval workflows, clauses that change by jurisdict

Meta, Walmart and Stripe publish the Personal Agent Protocol

Meta, Walmart and Stripe publish the Personal Agent Protocol

The agent economy is writing its rulebook tonight: one open standard for AI bots at the checkout, a cheaper image model from Google, and Anthropic turning bug-hunting into a tiered product. Meta, Walmart, Stripe and Sierra are publishing an open "personal agent protocol" — a standard that defines how personal AI agents interact with businesses online. The group behind it reads like a cross-section of agentic commerce: Meta, Sierra, Walmart, Stripe, Shopify, Genesys, Rocket, NiCE, Decagon and I