With the second beta of iOS 18.2 that’s available for developers today, Apple has further fleshed out the ChatGPT integration that’s available with Siri. In the Settings app, there’s now a section that shows the ChatGPT daily limit, and offers an option to upgrade to the paid ChatGPT Plus plan.
The beta includes an Advanced Capabilities section with a “Daily Limit” reading that shows up as “Under Limit” without a paid ChatGPT plan. Users have access to a small number of requests that use the most advanced ChatGPT capabilities, and requests are downgraded to a basic version of ChatGPT after that.
OpenAI has long restricted access to ChatGPT’s most advanced feature set to a paid Plus plan, so accessing ChatGPT through Siri is subject to the same limitations.
There is an option to upgrade to ChatGPT Plus, which is priced at $19.99 per month. ChatGPT Plus provides 5x more messages on the newest version of ChatGPT, GPT-4o. It also allows for higher limits on photo and file uploads, image generation, and web browsing, as well as an option to converse with ChatGPT using advanced voice mode.
Customers who already have a ChatGPT Plus plan can sign into their account, and those that do not can sign up through the Settings app. The Settings app opens to an in-app browser where users can sign up directly with OpenAI.
Free ChatGPT users won’t have access to OpenAI’s latest models (such as o1-preview) or premium features such as Advanced Voice Mode. They also can only make two images with Dall-E per day, and can’t send as many messages to the AI chatbot as premium users.Free access to ChatGPT-4o requests resets every 24 hours, and when the limited number of requests are used up, Siri will switch to a more cost effective version of ChatGPT.
It’s also unclear whether Apple is taking a cut of the revenue OpenAI generates from ChatGPT Plus signups through the Settings app. It’s possible the iPhone maker is simply betting that having cutting-edge AI features is worth the free exposure it’s giving OpenAI, because it will push enough customers to upgrade to new phones.
iOS 18.2 adds other new AI features, too, including Visual Intelligence and Genmoji. TThe update is set to arrive the week of December 2nd, Bloomberg’s Mark Gurman reports.
Given up on trying to get Google Assistant to dim the lights or pull your drapes? If you’re an Android user, now’s your chance to give Gemini a spin with your smart home.
Google has just released an extension for the Gemini app that allows Google’s LLM to take charge of a wide range of smart gadgets, from smart lights and thermostats to smart shades and TVs.
For now, you’ll need the Android version of the Gemini app, and you’ll also need to be enrolled in Google’s Public Preview program.
Once you’ve enabled the Google Home extension on the Gemini app (this Google support page has all the details), you’ll be able to give smart home commands to Gemini that had previously been executed by Google Assistant.
Google explains that telling Gemini the phrase “the sun is too bright in the living room” will result in the AI assistant closing the smart window blinds in your home. The company also gives an example of using “help me clean up the kitchen” to start a robot vacuum.
There are a few limitations to note, though. It’s only available on Android phones and is restricted to English prompts.For now, only the following categories are supported (according to Google):
Lights & power, like lights, outlets, and switches
Climate control, like air conditioning units, thermostats, heaters, and fans
Window coverings, like curtains, blinds, and shutters
Media devices, like TVs and speakers
Other smart devices, like washers, coffee makers, and vacuums
Google warns that Gemini’s smart-home control abilities are (for the moment, anyway) “for convenience only, not safety- or security-critical purposes.” On devices that don’t have Gemini, you’ll still be using Google Assistant.
Specifically, Gemini can’t do anything with a security device that requires a PIN, stream video feeds from security cams, or perform “some” actions on gates, locks, doors, garage doors, and other security-related devices.
You can try the Google Home extension in the Gemini app now if you’re enrolled in the early preview, and a wider release is likely to follow in the future.
According to Korean blogger yeux1122 (translated), Samsung is planning to launch a more affordable Galaxy Z Flip FE model next year. While the leaker didn’t provide further details on the name or specifications, they hinted that the Galaxy Z Flip FE might be released “simultaneously” with the next-generation Galaxy Z Flip, likely called the Galaxy Z Flip 7. This could mean that Samsung may introduce both models at the same time next year.
Moreover, according to a Samsung executive (spotted by Jukanlosreve on X), the company is “considering ways to lower entry barriers” for foldable phones “so that more customers can actually experience” them. This statement, coupled with the leak above, hints that a more affordable foldable model may be on the horizon. The executive also mentioned that Samsung is “preparing new form factors,” which could be a reference to the rumored Samsung tri-fold device.
There’s no word on what a Galaxy Z Flip FE would look like, but we’re guessing this model would offer an older flagship processor (if not a mid-range SoC), 128GB of base storage instead of 256GB, less RAM, and a less impressive camera setup.
Why a budget-friendly Z Flip FE makes sense for Samsung
There’s a strong argument that Samsung needs to release a more affordable foldable phone sooner rather than later. This comes after reports that the Galaxy Z Fold 6 and Z Flip 6 saw much lower pre-order figures than last year’s models. More recently, ZDNet Korea reported that Z6 sales forecasts have been revised from nine to 10 million units down to seven to eight million units. The outlet also noted at the time that Samsung was considering a “popular” version of its Flip phone.
There’s no word on pricing for this cheaper Z Flip model. However, the Motorola Razr 2024 launched at $699, so we wouldn’t be surprised if a future Z Flip FE model matches this price tag.
This isn’t the only leak from yeux1122. The Korean tipster also claimed that the recently leaked Galaxy S25 Slim could launch in April and suggested that the Galaxy Z Fold 7 could be accompanied by an additional Fold model. But we haven’t seen any significant details about this device.
In the week’s least surprising news, Amazon’s reinvention of its Alexa voice assistant has reportedly fallen even further behind. According to Bloomberg, the launch of a new Alexa — billed as a smarter, more capable AI-powered voice assistant — has been pushed back. Again. “A person familiar with the matter said Alexa AI teams were recently told that their target deadline had been moved into 2025,” writes Bloomberg.
In June, Fortune reported that the AI-powered Alexa — which Amazon demoed last September and said would be available for a free preview on its Alexa-fitted devices in the U.S. — is not even close to being ready. Former employees told the publication that the company doesn’t have enough data nor access to the chips needed to run the large language model (LLM) powering the new version of its voice assistant. The company also reportedly deprioritized Alexa AI to focus on building generative AI for its cloud computing unit, Amazon Web Services.
Amazon said its former employees are incorrect and uninformed on its current Alexa AI efforts, and that the Amazon Artificial General Intelligence team has access to both in-house Trainium chips and Nvidia’s graphics processing units (GPUs).
“We have already integrated generative AI into different components of Alexa, and are working hard on implementation at scale—in the over half a billion ambient, Alexa-enabled devices,” an Amazon spokesperson previously told Quartz in a statement. “We are excited about what we’re building and look forward to delivering it for our customers.”
On Amazon’s third-quarter earnings call, Jassy said Amazon has a “very deep partnership with Nvidia” and plans to “be their lead partner on most of their new chips.” Production of the second version of Amazon’s training chips, Trainium, will start ramping up in the next few weeks, Jassy said.
Bloomberg reports that Amazon CEO Andy Jassy has yet to convey a compelling vision for an AI-powered Alexa to the company. While he’s said publicly, “We continue to re-architect the brain of Alexa … ”, there’s been scant information about what an LLM-powered Alexa will bring to its millions of users — beyond being able to converse more naturally. More importantly, it seems Amazon has yet to prove it can do this without diminishing the features customers use the assistant for every day.
While the company searches for its vision, Jassy has installed a new head of the devices and services division under which Alexa falls. Panos Panay has been at the company for a year now, and Bloomberg reports the former head of Microsoft’s Surface division has “brought a focus on higher-quality design to a group adept at utilitarian gadgets.”
Amazon’s prior tact of making copious amounts of cheap hardware at the expense of better software is partly why Alexa hasn’t gotten measurably smarter over the last decade. However, with better hardware and a focus on building on Alexa’s strength, rather than simply turning it into a chatbot, the company could recapture Jeff Bezos’s original vision of creating Star Trek’s “Computer.” But whatever the plan is for a new Alexa, it looks like it won’t be here anytime soon.
At least one iPhone 17 model launching next year will be equipped with an Apple-designed Wi-Fi 7 chip, according to Apple supply chain analyst Ming-Chi Kuo.
All current iPhone models are equipped with a combined Wi-Fi and Bluetooth chip supplied by Broadcom, but Kuo expects Apple to equip “nearly all” of its products with its own in-house Wi-Fi chip “within about three years.”
The chip will replace the Broadcom chip that Apple currently uses for Wi-Fi and Bluetooth. Apple’s Wi-Fi 7 chip will be built using TSMC’s 7nm process. Apple apparently wants to move “nearly all” of its products to its own Wi-Fi chip within about three years, in order to cut supply costs and rake in more profits.
All four iPhone 16 models already support Wi-Fi 7 with Broadcom’s chip, but with some limited specifications. Kuo said Apple’s in-house Wi-Fi chip will support “the latest Wi-Fi 7 spec,” but he did not provide any further details. The chip will be manufactured with TSMC’s 7nm manufacturing process known as N7, he added.
Kuo has aligned with information shared last year by Jeff Pu, another analyst who covers companies within Apple’s supply chain. Pu said the iPhone 17 Pro models would be equipped with an Apple-designed Wi-Fi 7 chip, and he said the in-house chip would expand to the entire iPhone 18 lineup the following year.
Apple is also working on its own 5G chip, which will be separate from the Wi-Fi one and built using a different TSMC process technology. This will also start to be used gradually from next year onwards. The first product to come with Apple’s own 5G modem will be the next iPhone SE. This will however still use a Broadcom Wi-Fi chip. There have been conflicting rumors about whether the Apple-designed 5G and Wi-Fi chips will be separate chips, or one combined chip with 5G, Wi-Fi, Bluetooth, and GPS capabilities.
Wi-Fi 7 allows for data transmission over the 2.4GHz, 5GHz, and 6GHz bands simultaneously with a supported router, resulting in faster Wi-Fi speeds, lower latency, and more reliable connectivity. Wi-Fi 7 can provide peak speeds of over 40 Gbps, a 4× increase over Wi-Fi 6E, if a device supports the maximum specifications.
In a major shift of its release cycle, Google has revealed that Android 16 will be released in Q2 of 2025. Google typically pushes out a new major release of Android in Q3 or Q4, but the company has decided to move next year’s major release up by a few months so more devices will get the update sooner.
Google’s updates kicked off by highlighting the future of its Android SDK. In a blog post, the company states it will bring “more frequent” SDK releases, beginning in 2025. Google states it will have a “major” release in Q2 that contains “behavior changes” for apps. This seems to be Android 16 as Google states developers should keep their eyes out for more information regarding its first preview. The company states app developers can prepare for the major Q2 2025 release by conducting the “annual compatibility testing” a few months sooner.
The minor SDK release in Q4 2025 will contain feature updates, optimizations, and bug fixes to clean up any issues with the intended major Q2 launch.
Q1 and Q3 2025 will be minor, offering “feature only” updates to developers, per Google’s set timeline.
Gemini in Android Studio Updates
Following the debut of “Gemini in Android Studio” in April,the company is pushing the boundaries of its AI by boosting its coding capabilities, allowing it to edit your work by “implementing common design patterns.” Several of Gemini’s new code editing features include the ability to “modify and refractor” code based on custom prompts.
A commit message generation feature lets the AI analyze your changes and propose its own set of alterations. “Rethink and rename” lets Gemini generate names for your coding classes, methods, and variables. “Prompt Library” enables the AI to “automatically” save and manage a user’s most frequently used prompts for easy recall. Users can also generate documentation for a select chunk of code by right-clicking.
In the Android Studio, Gemini can now auto-generate “composable” previews and simplify the visualization process for your intended UI. Moreover, “coming soon,” Google says Gemini in Android Studio will support image attachments.
As Gemini in Android Studio picks up a Ladybug feature drop, developers can weed out bugs quicker and (hopefully) see fewer crashes. Additional features include test scenarios using local code context, build/sync error insights, and app quality insights for suggested crash fixes.
Play Store Updates
In a separate blog post, the company states it’s changing how apps are displayed on its store. These new formats include a full-screen mode for portrait videos in an app store listing. Google states a more “prominent” install button should now be visible to users.
Google highlights the Play Store’s “personalized” query recommendations. According to the post, users can find these recommended search queries via the “Search Home.” This is available for English with support for additional languages set to arrive “soon this year.”
Google’s “interest picker” arrived to help the Play Store understand where your interests are for accurate recommendations. The company rolled out a similar tool for games in July.
The post adds that the Play Store delivers a more robust security system when purchasing content. The company says users can now choose “on-device biometrics” when purchasing an app. This includes deciding between using your fingerprint or face to authenticate a purchase instead of inputting your account’s password.
The final two updates for the Play Store involve boosting revenue for developers and their apps alongside Play Integrity API upgrades. Concerning the former, Google states that it has “proactively” encouraged users to assign a payment method on the Play Store and within the Android OS. This has reportedly “doubled the number of purchases” in 2024 and provided a nearly 3% increase in conversion rate.
Payment options have continued to expand as users find options like Paypal, Cash App, direct carrier billing, and more. Additionally, Google highlights its Play Store algorithm for pushing what content users might buy as additional content for their favorite apps.
Google also highlighted its “gentle nudge,” an alert that reminds users of an item they’ve forgotten in their cart.
Google states, “in the coming months,” developers can opt into improved Play Integrity API verdicts to bolster their app’s defenses. What’s more, Google’s “app access risk” feature has exited beta. Developers can now “detect and respond” to apps that capture a user’s screen or control their device to avoid security disasters.
After debuting on the Fitbit Ace LTE, Google Wallet with tap-to-pay will be available next year for kids using Family Link-managed Android devices. The news comes after Google first added the feature to its kid-focused Fitbit Ace LTE fitness tracker in August.
Google tells us today that supervised kids in Family Link will be able to download the Google Wallet app on their Android phones. This will allow them to make tap-to-pay transactions in stores. They won’t, however, be able to use cards saved in Google Wallet for online payments.
Parents can supervise children’s accounts via the Google Family Link app that lets parents set restrictions for content and screen time, and added the ability to see when kids leave school and arrive home in 2022. Meanwhile, their Family Link app will let them view recent transaction history and remotely remove a card, as well as block passes.
Kids will be able to make tap-to-pay purchases using standard authentication options, including fingerprint, facial recognition, PIN, or passwords. They can also use Google Wallet to store gift cards and event tickets, but they can’t make online purchases or save health or identification cards.
Google already built a lot of this infrastructure out for its Wear OS-powered kids smartwatch earlier this year, noting the “the positive response of tap-to-pay on Fitbit Ace LTE.” The company emphasizes how this “new experience is built with safety in mind.”
Google’s kids’ payments feature will start rolling out to Google Wallet users in the US and several other countries next year.With the move, Google is catching up with Apple, whose Apple Cash Family service already lets children use Apple Pay for purchases in stores. It also lets kids buy things online as well as send and receive money through messages.
Apple introduced new Mac mini, iMac, and MacBook Pro models this week, adding faster and more efficient M4 chips, along with some other updates like Thunderbolt 5 and nano-texture display options for some models. One upgrade we thought we might see was support for the latest Wi-Fi specification, but the new machines did not get a Wi-Fi 7 upgrade.
All of the new M4 Macs continue to offer Wi-Fi 6E, and while it does allow for access to the 6GHz band on supported Wi-Fi 6E and Wi-Fi 7 routers, it’s not the latest and greatest technology.
Apple’s Wi-Fi specifications page confirms that the Wi-Fi 7 chips in the iPhone 16 models have a maximum physical layer data rate of 2400Mb/s and a maximum channel bandwidth of 160MHz, which is the same performance as the Wi-Fi 6E chips in the iPhone 15 Pro models and the M3 Mac models with Wi-Fi 6E on 6GHz networks (Apple’s chart hasn’t been updated for M4, but presumably the Wi-Fi 6E chips have not changed).
Wi-Fi 7 in the iPhone 16 models does up the maximum data rate when connected to 5GHz bands compared to the Wi-Fi 6E chips used in Macs and iPads, but Apple does not support the full 320MHz bandwidth with its Wi-Fi 7 implementation. Wi-Fi 7 offers up to 320MHz bandwidth to accommodate more devices, but Apple has limited bandwidth to 160MHz, so iPhone 16 models don’t take full advantage of Wi-Fi 7 despite having a Wi-Fi 7 chip. That limitation has been a disappointment, but the iPhone 16 models do support Multi-Link Operation (MLO) for connecting to multiple bands at the same time for faster data transmission and lower latency on supported networks.
Apple certainly could have opted for a Wi-Fi 7 chip without limitations in the M4 Mac models, especially for the more expensive M4 Pro and M4 Max machines, but that didn’t happen. Given that people often keep Macs for several years, it is curious that Apple opted not to offer this bit of future proofing in the Mac line.
Customers who plan on purchasing one of the new M4 Macs should know that these machines continue to offer Wi-Fi 6E features and lack the faster speeds and the latency benefits that come with Wi-Fi 7.
Apple is now letting developers try more Apple Intelligence features. The company has released the first developer beta of iOS 18.2, and it adds tools like the ability to generate emoji with Genmoji and images with Image Playground, more AI-powered writing features, integration with ChatGPT, Visual Intelligence to search with your camera on iPhone 16 phones, and more.
ChatGPT is an opt-in
Anxious users need to be reassured: ChatGPT doesn’t just “turn on” by default and no data is sent to OpenAI just by turning on your iPhone with Apple Intelligence (at least that’s what Apple promises). Instead, it is a separate switch in the system settings that must be explicitly activated – along with the extensive information dialog. “If the iPhone works with ChatGPT, it can do more for you,” it says.
Setting up ChatGPT in iOS 18.2
To begin using ChatGPT in iOS 18.2, you will need to become a member of the Apple developer beta program. Once you have done that, you will need to download and install the update. From there, enabling ChatGPT is simple, just go back to your Settings. Open Settings and go to Siri & Search.
Turn on Apple Intelligence and navigate to the ChatGPT toggle to activate the feature. Once ChatGPT is enabled, it’s accessible through Siri, allowing users to engage in intelligent conversations, get visual insights, generate images, and more without additional apps.
I was curious I should cancel my ChatGPT Plus subscription. Like, was ChatGPT + Siri somehow a better or more powerful version than ChatGPT Plus? The short answer is no; if you’re a ChatGPT Plus subscriber, simply link your OpenAI account in the settings section so you can still access the premium features.
OpenAI generates images via Siri
As mentioned, ChatGPT is addressed via Siri – this can be done by voice or “Text to Siri”. If you want, you can switch off Siri’s requests that something should go in the direction of ChatGPT. If you want to use ChatGPT a lot, you can do this, but you are putting yourself more in the hands of OpenAI. It is even possible to send all requests to ChatGPT by default by simply saying “Ask ChatGPT”. The image generator DALL-E 2 is also integrated. This means you can now use Siri to create AI images – Interestingly, even before Apple itself has released its Image Playground including Genmoji and image wall function, as many beta users are currently still on hold. ChatGPT and DALL-E 2 also generate photorealistic images, which Apple does not provide.
In the Writing Tools, ChatGPT now allows you to create completely new texts. Previously, Apple’s own tools could only rewrite existing content, “translate” it into other formats and check for errors. Now there is a “Compose” button that uses ChatGPT. In practice, it still takes a little longer for ChatGPT to respond. Apple has apparently not created a “fast lane” to the OpenAI servers as part of the developer beta. Incidentally, Apple is not planning to integrate ChatGPT (even) more deeply into the operating system – New Siri functions such as the use of the user context (e.g. emails, address book, appointments) will be reserved for Apple technology and will also largely run locally on the user’s own device when they are released in the coming months.
Google introduced Application-Bound (App-Bound) encryption in July (Chrome 127) as a new protection mechanism that encrypts cookies using a Windows service that runs with SYSTEM privileges.
The goal was to protect sensitive information from infostealer malware, which runs with the permissions of the logged user, making it impossible for it to decrypt stolen cookies without first gaining SYSTEM privileges and potentially raising alarms in security software.
“Because the App-Bound service is running with system privileges, attackers need to do more than just coax a user into running a malicious app,” explained Google in July.
“Now, the malware has to gain system privileges, or inject code into Chrome, something that legitimate software shouldn’t be doing.”
However, by September, multiple information stealers had found ways to bypass the new security feature and provide their cybercriminal customers the ability to once again steal and decrypt sensitive information from Google Chrome.
Google responded by saying that it was expected, and added that it was happy the changed forced a shift in attacker behavior.
“This matches the new behavior we have seen. We continue to work with OS and AV vendors to try and more reliably detect these new types of attacks, as well as continuing to iterate on hardening defenses to improve protection against infostealers for our users.”
Now, security researcher Alexander Hagenah built and shared a tool on GitHub he called ‘Chrome-App-Bound-Encryption-Decryption’ which does the same as these infostealers, BleepingComputer reports.
“This tool decrypts App-Bound encrypted keys stored in Chrome’s Local State file, using Chrome’s internal COM-based IElevator service,” the project page reads. “The tool provides a way to retrieve and decrypt these keys, which Chrome protects via App-Bound Encryption (ABE) to prevent unauthorized access to secure data like cookies (and potentially passwords and payment information in the future).”
Commenting on all of the above, Google essentially said it was satisfied, since crooks now need higher privileges to pull off the attacks:
“This code [xaitax’s] requires admin privileges, which shows that we’ve successfully elevated the amount of access required to successfully pull off this type of attack,” Google said.
