Apple Touts AI Intelligence, Inks ChatGPT Deal While Elon Musk Pouts

Getty Images

After months of speculation, Apple did what we all thought it was going to do at its annual developers' conference last week, when it announced a sweeping set of generative AI tools and functionality for the iPhone, iPad, Mac and Apple Watch.

CEO Tim Cook unveiled "Apple Intelligence," describing it as a "personal intelligence system" that understands the context of all your personal data so it can deliver "intelligence that's incredibly useful and relevant" and thus make your "devices even more useful and delightful."

That intelligence will allow iOS 18 and other operating system software to use your personal data — texts, emails, documents, photos, videos, audio, calendar, search history — to deliver answers and help you get things done.

That includes writing emails and rewriting emails in different styles; checking your grammar and word choice; and editing photos with a Clean Up tool that lets you remove things you don't like in the image, including people in the background. Apple says it will also help tidy up your email inbox with a feature called Priority Messages that will highlight the most urgent emails, like a boarding pass, CNET's Lisa Eadicicco reported.

There's a new gen AI emoji maker, called Genmoji, that lets you create custom emoji with a few simple words — smiley face with cucumber slices over its eyes — or turn photos of family and friends into emoji to share in your messages. There's an image generator called Image Playground that lets you pick from a palette of themes, accessories, places and more to use generative AI to create images for use in Messages, Apple's Freeform and Pages apps, and in a new Image Playground app.

A big part of Apple's strategy is built around giving a larger speaking part to its Siri voice assistant, a welcome update since it seems Apple had let Siri lag behind rivals like Alexa and Google Assistant over the years. Coming this fall, "You'll be able to speak in more natural language, and Siri should be able to understand you even if you stumble all over your words. Since it understands context, you can ask follow-up questions, and Siri should know what you mean," Eadicicco reported.

Siri is also getting on-screen awareness so it can take action based on what's on your phone's screen. That could be anything from adding someone's address to their contact card after getting the info in a text message to sharing photos on your behalf.

"If you ask it when a loved one's flight is landing, for example, it'll cross-reference flight data that may have been shared with you through a text message or email," Eadicicco said. "Siri will also be able to incorporate other information, like upcoming lunch reservations, so you can ask it how long it will take to get from the airport to the restaurant without switching between apps."

Of course, getting all that personalized context means giving Apple access to all your data, making privacy and security real concerns. Which is why the company said in its keynote, in its general press release, in a privacy press release and in a post on its security site that it's created a "new standard for privacy in AI."

Apple's new standard for AI security, as I explain in more detail in our privacy story last week, is about making sure your data is protected and secure, whether the ingesting, digesting and manipulation of all that data is done on your personal device (also known as on-device or local processing) or a complex AI task needs to be handed off to more powerful computer servers in the cloud running custom Apple chips.

Apple's promise, as part of its new Private Cloud Compute standard, is that, just like how it handles on-device processing, the company "uses your data only to fulfill your request, and never stores it, making sure it's never accessible to anyone, including Apple."

In addition to all that, Apple also signed a deal to give its users access to OpenAI's popular ChatGPT chatbot in Siri and in its Writing Tools as an optional feature. In addition to ChatGPT, Apple said it intends "to add support for other AI models in the future" but didn't say what those other tools might be.

The deal was enough to rile Elon Musk, a co-founder of OpenAI who's now creating his own rival to ChatGPT. More on that — and Musk's claims of security concerns — below.

If you take away one thing from Apple's announcements, it's that gen AI is coming to the masses, whether you think the technology is great or poses hazards (or both). Over a billion people use Apple's iPhone, and the company's track record demonstrates that Apple can help popularize and socialize technology, showing you new ways of doing things that you hadn't thought of before.

Noted Wired: "Other tech companies want to sell you chatbots. Apple's demos show the value of seeing the AI as an integrated, holistic experience rather than a stand-alone app or device."

Here are the other doings in AI worth your attention.

Apple's ChatGPT deal prompts Musk to ban employees from using Apple devices

Apple's AI announcements, the top news of the day on June 10, attracted the attention of investors, fans and critics — including billionaire entrepreneur Elon Musk.

Musk threatened to ban his employees and visitors from using Apple devices at the companies he runs, saying that he's no longer a fan of the iPhone, the iPad and Mac computers because he has security concerns about whether Apple's new partnership with OpenAI will protect users' personal data.

In posts on X, the social media network he owns, Musk alleged, without providing any specific details, that building OpenAI tech at the operating system level would be an "unacceptable security risk." He then later described the partnership as "creepy spyware" and said Apple was basically not smart enough to build its own chatbot and had "no clue" that OpenAI would deliver on its privacy promises to Apple users.

There's a lot to unpack there, and if you want to get into the nitty-gritty, my colleague Sareena Dayaram and I broke it all down for CNET readers last week.

But the bottom line is this: Musk, one of the world's richest men, who has a reputation for bluster, is the head of a startup developing a rival to ChatGPT called Grok and was a co-founder of OpenAI, a company he was suing until he dropped that suit last week.

So, his criticisms may be more complicated than just worries about security. Which is why he was being called out by members of his own social media fact-checking community, who were saying his remarks about the Apple-OpenAI partnership were inaccurate and misleading.

As for the OpenAI lawsuit, Musk, who's worked to portray himself as an advocate for users and humanity, also didn't mention his legal beef with OpenAI, which is detailed in his February lawsuit. In that lawsuit, he claimed that the San Francisco-based startup, led by CEO Sam Altman, abandoned its founding mission to develop AI that will benefit humanity and instead has focused on chasing profit.

In response, OpenAI challenged Musk's narrative in a lengthy blog post on its site on March 5, saying the billionaire investor was angry that his 2018 attempt to take over OpenAI was rebuffed. It included Musk's demand to become the CEO and majority shareholder so he could turn it into a "for-profit entity" himself. OpenAI shared Musk's emails to back up its story.

Musk's lawyers didn't say why he'd decided to drop the suit a day before a judge in San Francisco was set to consider whether the suit should be dismissed, according to CNN.

Like I said, a lot to unpack.

AI's financial rewards — and its costs

Despite the soap opera aspects of how OpenAI is run — CEO Sam Altman temporarily fired by his board but being reinstated a few days later, top executives exiting, employees writing open letters calling for more transparency in how it's handling AI safety concerns — the company is reportedly raking in the bucks.

That's the word from The Information, which said the San Francisco-startup more than doubled its annualized revenue, to $3.4 billion, the past six months or so, according to sources who cited Altman's comments to his team. "Annualized revenue — a measure of the past month's revenue multiplied by 12 — was $1.6 billion in late 2023 and about $1 billion last summer," The Information said, noting that the cash influx is a sign the company is fending off rivals and continuing to win over businesses and consumers who have embraced its conversational AI.

OpenAI continues to draw more traffic than its AI chatbot rivals, a position it's held pretty much since ChatGPT introduced us to gen AI in November 2022, according to traffic data from Similarweb.

There's a lot of money to be made, with market research firm Statista estimating that the global AI market will reach $184 billion in 2024 and more than double, to $415.6 billion, by 2027.

But at what cost does all that innovation come, since we know that AI is a data-intensive business that requires a lot of computer power and energy to run? In a recent report, Scientific American touched on the potential cost to Google from using AI to answer billions of search queries each year.

There's no simple number, because there's agreement that costs will decline as these systems get faster/better at answering questions and as the cost of hardware declines. Still, here are three takeaways from Scientific American's report that are worth noting:

First: "John Hennessy, chair of Google's parent company Alphabet, told Reuters last year that an exchange with a large language model could cost 10 times more than a traditional search," Scientific American wrote. "Analysts at Morgan Stanley estimated that if AI generated 50-word answers in response to 50 percent of queries, it could cost Google $6 billion per year."

Second: Google told the publication that "machine costs associated with [generative] AI responses have decreased 80 percent from when [the feature was] first introduced in [Google's feature incubator] Labs, driven by hardware, engineering and technical breakthroughs."

And third, and perhaps most importantly, there's another cost that Scientific American makes clear — the cost to the environment: "Data centers, including those that house AI servers, currently represent about 1.5 percent of global energy usage but are projected to double by 2026, at which point they may collectively use as much power as the country of Japan does today. Generative AI itself is expected to consume 10 times more energy in 2026 than it did in 2023."

LinkedIn using AI to help job hunters

LinkedIn said last week that it's offering premium subscribers AI tools to streamline their job searches, after testing them out on users over the last year.

"Now you can look for jobs using intuitive and conversational language — such as 'Find digital marketing jobs in San Francisco that pay more than $100,000' — instead of solely using traditional search filters like job title or location," Rohan Rajiv, a product lead for LinkedIn Talent Solutions, said in a blog post.

The new premium AI tools also include AI-powered resume feedback and cover letter drafts. A premium subscription starts at $29.99 a month.

The company also said it's adding some new AI functionality for all LinkedIn users, including gen AI tools that will upload your resume and then draft suggestions for filling in sections of your profile.

Ukraine says AI-powered drones could be used for targeted assassinations

Ukraine, which unveiled the first AI-generated government spokesperson to the world in May, spoke at a NATO event in Poland earlier this month about how a drone, outfitted with enough sensors and AI to identify a person by the sound of their voice, could "take out a Russian war criminal with a targeted assassination," according to Bloomberg.

Though Ukraine's deputy tech minister, Alex Bornyakov, said the military drone was only in the "prototyping phrase," the AI technology needed to make it a reality exists today. Bloomberg reported Bornyakov as saying, "Computer vision works. It's already proven."

What's underlying all this is NATO's ethical framework for how AI can be used in warfare. David van Weel, a NATO assistant secretary general, told the gathering that it works with partners to "ensure reasonable human input in any lethal use of force." One way NATO is using AI is to read satellite footage to identify Russian aircraft, he said, according to Bloomberg.

Now the question is whether that ethical framework can be "codified into legally-binding rules, including by the United Nations," Bloomberg said.

I hope so. NATO expects to share an update to its AI strategy at the 2024 NATO Summit in Washington, DC, on July 9-11.

Human Rights Watch says personal photos of children wound up in AI database  

A report last week by Human Rights Watch reminds us why it's important that makers of large language models, or LLMs, share what training data has gone into creating their AI systems.

HRW said the personal photos of Brazilian children, including personally identifiable information such as their names, and when and where the child was when the photo was taken, have been found in database data used to train popular AI tools.

HRW said it found 170 photos of children from at least 10 Brazilian states, adding that, "This is likely to be a significant undercount of the total amount of children's personal data that exists in LAION-5B, as Human Rights Watch reviewed less than 0.0001 percent of the 5.85 billion images and captions contained in the data set."

The report noted that people might not realize how much information is scraped off the internet. "Many of these photos were originally seen by few people and appear to have previously had a measure of privacy. They do not appear to be otherwise possible to find through an online search. Some of these photos were posted by children, their parents, or their family on personal blogs and photo- and video-sharing sites. Some were uploaded years or even a decade before LAION-5B was created."

The photos, the organization said, can be used to generate explicit images of children from innocuous photos, as well as from images of child survivors of sexual abuse.

HRW is working with the German nonprofit that manages the LAION-5B database to have the images it identified removed. But the takeaway here is that we should all pay attention to whatever photos we're posting and sharing publicly.

Said HRW: "Fabricated media have always existed, but they required time, resources, and specialized expertise to create and were largely not very realistic. Today's AI tools create lifelike outputs in seconds, are often free, and are easy to use, risking the proliferation of nonconsensual deepfakes that could recirculate online for a lifetime and inflict lasting harm."

Sigh.

Elephants may have names for each other 

Let's end this week's AI news on a happy note.

Researchers who've been using AI to analyze voice recordings of elephant calls say they've found evidence that the animals, like humans, call each other by their own personal names, according to The New York Times.

Mickey Pardo, An acoustic biologist at the Cornell Lab of Ornithology, published a study in the journal Nature Ecology & Evolution entitled "African elephants address one another with individually-specific name-like calls." These name calls differ from elephants trumpeting, which might "be their most recognizable sounds" but are in fact, Pardo said, "basically an emotional outburst."

He said the lower-pitched rumbles, captured by researchers as part of 469 vocalizations by family groups of adult elephant females and their offspring, are more meaningful, since they make up most of what the elephants are saying to each other. Said Pardo, "A lot of interesting stuff is going on in the rumbles."

I think that should be a slogan for a T-shirt.

Added the NYT: "The scientists are not sure precisely which part of a vocalization might be the elephant's name. But they found that their AI tools' ability to identify the intended recipient of a rumble far exceeded what random chance would dictate." The elephants would perk up their ears and rumble back when they heard what the researchers believe to be their names.

George Wittemyer, who is chairman of the scientific board for the nonprofit Save the Elephants, called the Cornell research a "game changer." Wittemyer told the NYT that it shows how important "that social fabric is to the every existence of this animal. Social bonding is fundamental to everything about elephants."

Editors' note: CNET used an AI engine to help create several dozen stories, which are labeled accordingly. The note you're reading is attached to articles that deal substantively with the topic of AI but are created entirely by our expert editors and writers. For more, see our AI policy.