The Vergecast
The Vergecast

The real story of the iPhone 18 Pro's camera

This week's iPhone 18 Pro discourse made aperture a priority, but today on Vergecast we zoom out to talk about the state of smart phone photography at large. Executive editor Jake Kastrenakes, guest host David Imel and senior reviewer Allison Johnson get into manual controls, "buckets of p

Featured Speakers

Vox Media Podcast Network Host

Topics Discussed

Episode Summary

Executive Summary: The episode opens with sponsor spots, then spends most of its time on two deep-dive conversations: a highly technical review of the iPhone 18 Pro camera upgrades (variable aperture, improved image pipeline, Pro controls, photographic styles, and Apple’s new reference-image anti-tampering feature), followed by a discussion of the escalating AI safety debate after high-profile researcher warnings, cyber incidents, and new efforts by AI labs to add third-party evaluators. It closes with a lightning round about autonomous vehicle monitoring, e-readers, and quirky new cameras.

Main Topics: iPhone 18 Pro camera upgrades and real-world performance (Priority: 5/5): Allison Johnson and David discuss Apple’s new camera hardware/software changes, especially the variable aperture, better low-light behavior, improved sharpening/color, and expanded pro controls. They agree the most meaningful gains are not just the headline aperture feature but the cleaner image pipeline and more usable manual controls. Apple’s new photographic styles and camera UX (Priority: 4/5): The conversation examines Apple’s expanded style controls, including grain, texture, tone mapping, and palette adjustments. The hosts note that Apple is making complex creative controls accessible, but the menus and naming can feel overwhelming. Apple’s reference-image anti-deepfake feature (Priority: 4/5): The hosts unpack Apple’s new reference image mode, which cryptographically signs images at the sensor level so users can later view an unedited 'negative' version. They see it as a promising but narrow, voluntary tool for photographers and photojournalists rather than an everyday solution. The AI safety backlash and 'race to superintelligence' (Priority: 5/5): Hayden Field explains why recent warnings from a former Anthropic/OpenAI researcher resonated, linking them to recent AI cyber incidents, fears about rogue agent behavior, and broader public concern that labs are losing control of their systems. Accountability, audits, and lab-controlled safety (Priority: 5/5): The discussion covers Dario Amodei’s call to ‘pace the frontier,’ with third-party auditors/embedded evaluators inside labs, but questions whether voluntary audits can ever create real accountability when the companies themselves control scope and access. Lightning round tech oddities: Waymo monitoring, Books Palma, and tiny cameras (Priority: 3/5): In the fast segment, the hosts discuss Waymo’s apparent ability to detect firearms and contact police, a new premium e-reader race led by Books Palma 3, and two compact retro-style cameras from Camp Snap and Fujifilm.

Key Arguments: Apple’s most important camera improvements are software and processing changes, not just the variable aperture; the new image pipeline produces better color and less over-sharpening. Manual controls in the iPhone camera app are now actually usable, with quick access to shutter speed, white balance, and long exposure settings without digging through the Settings app. Apple’s reference-image mode is a smarter anti-tampering approach than metadata-only standards because it is tied to the secure hardware and cannot be bypassed by a screenshot. The AI safety panic feels new to the broader public because recent cyber incidents made the risk of autonomous agent behavior concrete, not because the underlying concerns are novel. AI labs have a built-in incentive to publicize power and danger simultaneously: fear helps marketing and fundraising, but uncontrolled behavior threatens enterprise customers and trust. Third-party AI audits are useful only if evaluators have meaningful access; otherwise, audits become pre-scripted public relations exercises. Voluntary lab commitments to 'pace the frontier' resemble a prisoner’s dilemma: if one company slows and another doesn’t, the race just continues. In autonomous vehicles, users need clearer disclosure that in-car monitoring may be used to detect violations and involve law enforcement. Hardware still matters most for phone cameras; if Apple wanted a big leap, a larger sensor or better lens system would matter more than additional software polish.

Data Points: Trusted by AI legal teams: More than 60% of the Am Law 100 - Harvey AI sponsor spot Proposed manual exposure control: Up to 1 second of long exposure - iPhone 18 Pro Pro Controls discussion Camera sensor size: 48 megapixels - Apple’s main sensor and ProRaw discussion Reference image file size: Doubles the file size - Apple’s reference-image mode explained Third-party audit access window: 6 days - OpenAI’s limited outside audit after a cyber incident Audit respondents: 3 independent auditors - OpenAI’s transparency response after criticism Predetermined audit questions: 7 questions - OpenAI-approved scope for the independent audit Audit access extension: 8 weeks - Dario Amodei’s proposed evaluator access at Anthropic Audience of legal AI product: 60%+ of AmLaw 100 and leading Fortune 500 teams - Harvey AI sponsor spot Phone number sharing feature: One phone number for multiple iPhones - Discussion of iPhone Handoff / Duo ecosystem strategy Books Palma 3 price: $339 - New premium e-reader pricing Books Palma 2 price: $279 - Prior-generation e-reader pricing Books Palma 3 storage: 128 GB - New e-reader hardware specs Books Palma 3 battery: 3,950 mAh - New e-reader hardware specs

Pivotal Quotes: "The future of law is agentic." — Harvey AI ad: Opening sponsor spot describing AI agents for legal work "You can only access through third-party camera apps like Mo. Or Hallide." — David: Explaining the difference between RAW and ProRaw on iPhone cameras "We’re the good guys and we have to beat these other irresponsible people, it’s a little bit less compelling." — Beth Barnes (quoted by Hayden Field): Summarizing the skepticism toward lab-led self-regulation in AI safety

Implications: The episode suggests camera hardware is still bounded by physics, while AI governance is still bounded by incentives: users may get better tools, but meaningful trust will depend on transparency, not just features or promises.

🔓 Sign Up for Unlimited Episode Search

About The Vergecast

The Vergecast is the flagship podcast from The Verge about small gadgets, Big Tech, and everything in between. Every Friday, hosts Nilay Patel and David Pierce hang out and make sense of the week’s most important technology news. And every Tuesday, David leads a selection of The Verge’s expert staffers in an exploration of how gadgets and software affect our lives – and which ones you should bring into yours.

View all episodes from The Vergecast