The iPhone Duo got all the attention, but Apple also used yesterday's event to debut Apple Watch and iPhone features that could put its mark on the AI era.

Why it matters: Apple is aiming to show it can be more private and useful, even as it plays catch-up to the state of the art in AI.

Catch up quick: Apple's event was largely focused on hardware, from the $1,999 foldable iPhone Duo to the iPhone 18 Max to new Apple Watches and AirPods.

• Most of the products will ship on Sept. 18, but the Duo isn't coming until October.
• The most notable thing on the traditional phone front might have been the lack of a new base-level iPhone. Instead, Apple is keeping the iPhone 17, iPhone 17e and iPhone Air around, albeit with some price hikes to account for higher component costs.

Zoom in: Apple spent a significant amount of time between and around the hardware announcements to talk about Apple Intelligence, the revamped Siri and other AI features.

• Live Rewind allows you to quickly catch up on or revisit what was just said by showing a text snippet on your Apple Watch.
• Siri Recap takes notes on your conversations throughout the day and creates high-level Apple Intelligence-generated summaries. You can stay present in the moment and use the summaries later to jog your memory. Your summaries auto-delete from your watch after seven days if you choose not to save them.
• Sound Recognition alerts Apple Watch wearers to specific sounds, such as a doorbell or siren. These notifications are particularly useful for those who are deaf or hard of hearing.

Between the lines: In an era where plenty of tech companies are offering AI transcription gadgets that record entire conversations, Apple is making the case for a different approach.

• Rather than store recorded audio, Apple is transcribing what its microphones pick up, and providing certain highlights back to the user without attribution to specific speakers. Highlights can be the just-recorded 15 seconds or a daily recap.
• IDC group VP Tom Mainelli said in a statement that Apple is "attempting to walk a fine line between always-listening ambient AI computing and its desire to respect [the] user's privacy."

Separately, Apple debuted Apple Reference Image — a feature in the iPhone 18 Pro that records verifiable proof that an image hasn't been altered (by AI or otherwise).

• When a photo is taken in the new "reference mode," the iPhone camera captures signed sensor data that is turned into an unalterable reference image, akin to a digital negative.

What they're saying: Hany Farid, a longtime expert in deepfakes and computational photography, told Axios that the Apple Reference Image approach will take time to result in significant adoption because people both need new hardware and to choose to take a photo that includes the added data.

• At the same time, he said Apple's efforts could help spur further work in the field.
• "The symbolism of a company like Apple tackling the problem of trust in the visual record is significant," Farid said. "The problem of fake content is no longer a minor issue when one of the largest tech companies in the world decides to get into the game."

Anthropic researcher Jacob Coxon quit his job due to concerns about the safety of AI two months before his equity would have vested, he told Axios.

Why it matters: The disclosure raises the stakes on Coxon's now mega-viral resignation from the AI lab, which laid out the broad view that the tech could end humanity.

What they're saying: "I no longer have anything to gain by juicing up Anthropic's valuation. ... I left before any of my equity vested," Coxon said in an interview with Axios yesterday.

• In his post on X announcing his resignation, which now has over 145 million views, he wrote: "The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt."
• Other AI researchers, particularly from Google, have resigned over safety. But they did so after years of work, and presumably after their stock vested.
• Coxon was at Anthropic for just four months, and employees have to be there for six months for their stock to vest, he said. He still has equity in his prior employer, OpenAI.

The big picture: Anthropic has traditionally been viewed as more publicly cautious and safety-oriented than other frontier AI labs, making Coxon's departure particularly striking.

• Coxon said he has not seen Anthropic compromise safety to outlast its competitors, but he is concerned about the future: "If you're under pressure to race, you have to cut corners" or "skip steps in the oversight process."
• Those concerns can sometimes go too far, he said, describing what "sometimes feels like there's maybe excessive paranoia of OpenAI, excessive paranoia of China" that can help justify pushing ahead.

Threat level: As AI models are getting better, faster, they're also becoming harder to monitor.

• That, combined with the pressure to win on AI, was Coxon's breaking point that led him to quit.
• What was once science fiction about models knowing they are being tested, for example, is now "just a daily fact of working with these AIs," he said.
• "They know when they're being tested, and they will think about the fact that they're being tested."
• "The word doom is kind of silly," he said, adding that serious leaders in the industry are "on record saying they ... expect the potential for human extinction."
• The danger was evident during the many recent cyber incidents across frontier AI labs, he said.

The bottom line: Coxon is one of now several people who have seen AI's peak capabilities up close — and are warning that what they saw could end humanity as we know it.

Politicians have long made it hard to trust their words. Now, AI-generated campaign ads flooding the airwaves and social media are making it harder to trust your own eyes and ears.

Why it matters: Americans' trust in politicians and institutions has been tanking in recent years — and AI campaign ads could deepen that erosion, political tech experts say.

What they're saying: "I think that it is true that Americans are losing faith in elite institutions to tell them what is true and what is not, and AI exacerbates that trend," Nate Persily, co-director of the Stanford Law AI Initiative, told Axios.

• "The more artificial media that we consume, the less likely that people will be able to discern what is true from what is false, and also the less that they might care. And that's an important caveat."

Driving the news: Texas Senate nominee Ken Paxton recently released a realistic campaign ad targeting Democratic nominee James Talarico that used an AI-generated Talarico to deliver some of his most controversial statements.

• The synthetic Talarico combined and repackaged remarks Talarico had made about biological sex, God and Christianity, sometimes stripping them of their original context.
• The ad was based on real remarks, but it altered some of Talarico's wording while using an AI-generated version of him to deliver them.

Between the lines: Paxton's ad lands amid a broader backlash against AI, including increasingly visible fights over data centers.

• "The one thing that both right and left seem to agree on is that they don't trust AI," Sarah Kreps, the director of Cornell's Tech Policy Institute, told Axios. "They want authenticity, not AI-generated slop."
• Kreps noted Paxton may have run an unnecessary risk: "He'd probably be better served just letting the quotes speak for themselves, rather than making it a murky thing where he's risking conflating himself with all this AI backlash."

Zoom in: AI can degrade public trust even when people aren't fooled.

• "The main harm from the rise of AI-constructed media is not that people will be duped by something that is AI generated," Persily said. The bigger worry, he says, is that bad actors can dismiss something real as another deepfake.
• Even when voters recognize an ad as synthetic, the emotional impact can persist, UC Irvine epistemologist Cailin O'Connor said. "If you see some public figure saying something bad or doing something bad, you might still have negative emotions about them as a result."
• Kreps said persistent skepticism and vigilance can also become "tiring because it's just more of an inconvenience," leading people to assume content is fake until proven otherwise.

Zoom out: Not all political AI is equally deceptive.

• For example, cartoonish ads for Los Angeles mayoral candidate Spencer Pratt operated more like memes than deepfakes.
• Federal legislation is essentially nonexistent, while many states require deepfake disclosures for potentially deceptive ads and impose looser restrictions on clearly satirical content.
• In New York, state Democrats recently asked election officials to investigate a Republican gubernatorial campaign's AI ad over allegedly missing "materially deceptive" labels. The Republican says it was satire.

The bottom line: Without clearer rules, voters face a patchwork of AI campaign standards that could push them away from shared facts and further erode trust in politicians.

Go deeper: An explosion of AI deepfakes is redefining American elections

• Sen. Bernie Sanders (I-Vt.) will hold a bipartisan briefing next week to discuss the "extraordinary dangers" posed by AI, the latest scoop by our colleague Maria Curi. (Axios)
• China hit back at U.S. claims of "malicious" AI distillation ahead of planned talks between President Trump and President Xi Jinping. (Washington Post)
• How could AI actually kill us all? (Axios)

Mady again and I'm sure this is obvious to some, but I'm amazed by how much people are using AI while traveling. I see people using it in stores to ask if they're being charged a fair price, to translate menus, to ask for recommendations and so much more.

• I made friends with a girly who was using Chat to negotiate the price of a bag at a market. Bargaining can cross language and technical barriers!