When 23-year old Lindsey Brooke Isaacs surrendered to police at the advice of her attorney, she was treated like a hardened criminal. Facing eight felony charges, including three counts of vehicular homicide, she spent her first days in custody in solitary confinement, separated from the rest of the jail population.
Having been denied bond, Isaacs would spend nearly two weeks in jail, swallowing the prospect of spending the rest of her life behind similar bars.
“I was devastated, I was scared,” Isaacs told Fox News. “I thought my life was over.”
As grim as it looked, the Florida woman did have one thing going for her: she was completely innocent.
Isaacs’ arrest was first reported by WESH2 in June, but resurfaced by Fox News, which notes that the case around Isaacs proceeded solely based on evidence gathered from an automatic license plate reader (ALPR) provided by the extremely controversial company Flock Safety. Her charges stemmed from an incident in October 2025, when a horrific crash in DeBary, Florida killed three people, including a county official and his wife. One driver involved had fled, leading police to conclude that they were behind the tragic accident.
Using the Flock’s network of ALPRs, investigators canvased hours of surveillance footage recorded near the incident, following a tip by witnesses. The fleeing vehicle had been identified as a Dodge Durango — the exact kind of vehicle Isaac drove.
Sure enough, Fox News reports Isaacs’ Durango was traveling in the area at the time. A Flock ALPR had identified her vehicle three miles away from the crash site two miles prior to the incident, moving in the direction of the fatal crash. But that sighting is where the coincidences end, and the life-altering guesswork began.
During the investigation, witnesses and residue left on the impacted vehicle both pointed to a maroon-colored vehicle. On top of sporting a black paintjob, Isaacs’ Durango bore no damage indicating it had even scratched another vehicle per a Florida Highway Patrol analysis, let alone hit one hard enough to kill three people.
“Where is the damage on the vehicle?” the woman told Fox News, recalling her discussion with police, who impounded her car anyway. “Because I’m standing right in front of it and I don’t see any damage.”
Half a year later, Isaacs received the call from her attorney, informing her of the charges and advising her to surrender to police. After a 13-day stint in jail, she was released, after the Florida attorney general declined to bring any of the eight charges police had attempted to stick her with. Another woman was later identified as the driver of the maroon SUV and arrested on nine felony charges.
For her part, Isaacs says she’s filed a federal civil rights suit against two Florida Highway Patrol officers, claiming false arrest, wrongful imprisonment, and malicious prosecution.
“This took and is going to take maybe years to get better mentally, physically, emotionally,” she told Fox News of the ordeal. While the director of the Florida Highway Patrol extended a personal apology, it’s an insufficient gesture in the face of obviously shoddy police work and abuse of the ALPR network.
Even for the wrongfully accused, a two-week jail term is difficult to come back from. If Isaacs had a job she likely lost it, and there’s no telling what kind of reputational harm the ordeal will have on her career or her relationship with her community. That’s to say nothing of the psychological scars that come along with wrongful imprisonment, let alone a holding period in solitary confinement.
And though Flock was the technical lynchpin in this case, Isaacs told Fox News she holds more blame for police officers eager to snag a conviction, regardless of whether the accused was innocent or guilty.
The sad fact is that police in the US have long behaved this way, and there’s no reason to believe they’ll stop on their own. But now with unfettered access to Flock’s network of over 100,000 ALPRs, it’s becoming remarkably easy for officers to sculpt a criminal to fit their crime — inculpatory evidence be damned.
More on Flock: Trump Throws His Weight Behind Flock Cameras
The post Police Use Flock to Jail Completely Innocent Woman for Weeks appeared first on Futurism.
A new section in the iOS 27 privacy policy shows that Apple has reversed its stance on using customers’ data to train its AI models. Previously, the company assured customers that it had no need for customer data, and would figure out how to make Siri AI work well without it.
Apple maintains that it “takes steps to ensure that personal data is not stored or used by Apple.” But as Heise points out, and Apple has said in the terms of service for a few years now, the audio material may be listened to by Apple employees after anonymization.
[…]
It also admits that data collected may be reviewed by “review personnel.” It is unclear if “review personnel” are Apple employees or a third-party company contracted to read conversation transcripts, as it has historically been.
Starting today, when a user enables Siri AI, they will see a setup screen asking whether they want to opt in to “improve Siri and Apple Intelligence,” with two options: “Share Audio and Text” or “Not Now.”
[…]
What’s important to note here is that nothing is changing about the type of data Apple collects if you opt in to this setting. What’s changing is that Apple will now use the data to train and refine its Foundation Models, as well as Siri and Dictation like before.
iOS 26 had a toggle you could select to disable all Apple Intelligence features.
That toggle was removed for iOS 27. Every AI feature has to be turned off individually.
Previously:
Siri AI is beginning its public rollout as Apple transforms its familiar assistant into a substantially more conversational system built around personal context, onscreen awareness, web knowledge, and the ability to perform actions across apps.
The beta is available first in English on compatible devices running iOS 27, iPadOS 27, macOS 27, watchOS 27, and visionOS 27. French, Japanese, Korean, Portuguese, and Spanish support is scheduled for October. Availability remains more limited in some regions, with Siri AI initially unavailable on iPhone, iPad, and Apple Watch in the European Union and unavailable in China while Apple addresses regulatory requirements.
The release follows several years of gradual Apple Intelligence expansion, but Siri AI represents a more fundamental change. Instead of handling each request largely as an isolated command, the assistant can maintain conversations, find relevant personal information, understand what is displayed on a device, and combine those capabilities to complete multistep tasks.
Personal context is one of the biggest differences between Siri AI and previous versions of the assistant.
The system can search across a user’s messages, email, photos, notes, and other information when responding to a request. Apple offers an example in which someone asks what a family member suggested doing during an upcoming visit. Siri can locate the relevant conversation, find a recipe previously sent by email, and then place the ingredients into a Reminders grocery list.
That combination of retrieval and action is central to the new design. Siri AI can draft an email, edit and share photos, create calendar events, manage reminders, control media, and perform other supported actions without requiring users to manually move information between applications.
Third-party apps can participate through Apple’s App Intents architecture. Existing integrations include tasks such as messaging through WhatsApp or playing an audiobook in Audible, with Apple also previewing deeper capabilities in applications including Microsoft Outlook, Notability, and Tripsy.
Onscreen awareness adds another layer. Siri can identify information currently visible on a device and act on it. Someone viewing a sports website, for example, can ask Siri to identify upcoming home games and add them to Calendar without manually copying dates.
Apple has also expanded Siri beyond voice. On iPhone, users can invoke the assistant with the side button, voice command, or by swiping down from Dynamic Island. On iPad and Mac, Siri AI is integrated into Spotlight, while systemwide context menus can send selected files, images, or text directly to the assistant.
Siri AI is no longer confined to a temporary interface that disappears once a request is complete.
Apple has introduced a dedicated Siri app where users can begin new conversations or revisit previous ones. Conversation history can sync privately through iCloud across iPhone, iPad, Mac, Apple Watch, and Apple Vision Pro, allowing a discussion started on one device to continue on another.
Follow-up questions can retain the context of an earlier answer, giving Siri a more conventional conversational interface for longer exchanges.
The assistant also has access to broad web knowledge for questions that cannot be answered through personal information or local device data. Apple says Siri can retrieve current information for topics ranging from sports results to travel planning and nearby businesses.
Visual Intelligence expands alongside the assistant. On iPhone, a dedicated Siri mode inside Camera can identify and respond to objects or information in front of the user. A concert poster, for example, can become the starting point for questions about performance times followed by the creation of Calendar events.
On iPad, Visual Intelligence is integrated with screenshots. Mac users can select content on their displays through a keyboard shortcut, while Vision Pro users can ask Siri about app windows or physical objects simply by looking at them.
Writing Tools are similarly integrated across the system. Siri can generate drafts, revise existing text from natural-language instructions, and provide writing suggestions. In Mail and Messages, Apple says it can take a user’s typical communication style with a particular recipient into account, including tone and punctuation patterns.
Siri AI is powered by a new generation of Apple Foundation Models developed in collaboration with Google using Gemini technology, marking a notable shift in Apple’s AI infrastructure.
The architecture divides processing between models running directly on the device and larger models operating through Private Cloud Compute. Apple says requests sent to Private Cloud Compute are processed without personal data being stored or made accessible to the company.
Core resources such as the Spotlight index and Apple’s App Toolbox can remain on-device, allowing Siri to retrieve local information and perform supported actions without sending every request to a remote server.
More capable devices can also use AFM Core Advanced, Apple’s highest-performance on-device model. It powers improvements including more expressive Siri voices and advanced systemwide dictation that automatically handles capitalization, punctuation, and formatting.
Compatibility varies by feature. Apple Intelligence supports iPhone 16 models and later, iPhone 15 Pro and Pro Max, M1 or later iPads and Macs, iPad mini with A17 Pro, Apple Vision Pro, and supported Apple Watch models when paired with a nearby Apple Intelligence-enabled iPhone.
AFM Core Advanced requires newer hardware, including iPhone Duo, the iPhone 18 Pro line, iPhone Air, iPhone 17 Pro models, qualifying M4 or later iPads with at least 12GB of unified memory, M3 or later Macs with at least 12GB, and Apple Vision Pro with M5.
The initial Siri AI release remains explicitly labeled beta, and some of the surrounding Apple Intelligence features are scheduled for later updates.
Apple also confirms that features relying on server-side models can be subject to daily usage limits. That includes Siri AI, cloud-based photo editing, Image Playground, and AFM 3 Cloud models used through Shortcuts. Limits can vary depending on the feature, request complexity, system demand, and other factors.
Apple says expanded access to server-dependent features will eventually be available for a fee, introducing a future paid layer to parts of Apple Intelligence even as the core system remains integrated into supported devices.
Apple will also expand its intelligence capabilities on Apple Watch later in 2026. Planned Audio Intelligence features on Apple Watch Series 12 and Ultra 4 include Live Rewind, which can display text representing the previous 15 seconds of a conversation, and Siri Recap, which generates high-level summaries that can be reviewed later in the Siri app.
Those functions are designed around a hardware-isolated Secure Exclave on the S11 chip, with Apple saying raw audio is processed in isolation and immediately deleted rather than stored as recordings. Live Rewind and Siri Recap are scheduled to enter beta later in 2026, making the current Siri AI launch the beginning of a staged expansion rather than the final form of Apple’s new assistant.
The post Siri AI arrives with personal context, app actions, and a new privacy architecture appeared first on AppleMagazine.
The public messaging surrounding Meta’s flashy smart glasses hasn’t exactly gone in the company’s favor.
The camera-equipped devices have garnered a reputation for allowing shady individuals to record strangers without their consent, garnering them the memorable pejorative of “pervert glasses.”
So when Wired revealed in June that Meta had silently added facial recognition code to the smart glasses — stoking fears that the creepy glasses could be used to identify people on the street based on biometric data — that simmering distaste grew into a major PR disaster for the company.
Meta has ever since vehemently denied that it was planning on ever releasing the tech, a feature dubbed “NameTag.” That’s despite its CTO Andrew Bosworth happily describing those plans in excruciating detail in public and arguing it would be a “great feature” — mere days after calling Wired‘s bombshell reporting “absolutely dishonest.”
Now, the company is facing a proposed class action lawsuit over the ordeal, as Wired reports, with a set of families in Illinois and California accusing the company of illegally scraping their Facebook and Instagram photos to build a massive facial recognition system for “NameTag.”
The news comes just days after a mother in Utah accused Meta of using its AI to scour Instagram and Facebook posts to reveal invasive information about her young daughters.
Meta told Wired in a statement that the latest “lawsuit is without merit and misrepresents our work.”
“We’ve been transparent about how we use people’s information to build and improve our AI products,” the a spokesperson told Wired. “As for NameTags, nothing has shipped to consumers and no final decision has been made on what to do here, if anything.”
The proposed class action accuses Meta of extracting biometric signatures from publicly available social media profiles. As Wired first noted in June, these “faceprints” appeared in the code for the AI companion app of its smart glasses.
A patent described the software as matching these signatures to profile photos. An AI image generator Meta released called Emu was also trained on a vast swath of user-uploaded images and text. The complainants accuse Meta of using this AI system to harvest biometric data.
Meta has already been caught with its pants down after allowing users to generate images inspired by public Instagram accounts, admitting that it had “missed the mark” with the feature.
The company has a long history of mishandling sensitive user data, settling a class action lawsuit in 2020 over an entirely separate facial recognition system for $650 million. In 2024, it agreed to pay $1.4 billion to settle yet another facial recognition data lawsuit in Texas.
More on Meta’s glasses: Mother Horrified When Meta’s AI Scours Instagram and Facebook for Invasive Information About Her Young Daughters
The post Meta Is Using Instagram Users’ Photos to Build a Universal Facial Recognition System for Its Hated Smart Glasses, Class Action Lawsuit Claims appeared first on Futurism.
President Donald Trump continues to push the boundaries of what it means to be out-of-touch after he gave a ringing endorsement for Flock Safety, the deeply unpopular surveillance company behind a national network of AI-powered cameras and license plate scanners.
His reasons for backing the firm, however, weren’t particularly illuminating,
“I sort of like them because of that, because of law enforcement, but some people don’t,” Trump, told reporters on Sunday, sagaciously. “They think it’s an infringement,” he added, apparently speaking of Flock’s critics. “I like them.”
The comments are the first from Trump publicly endorsing Flock, according to the Washington Post. As recently as last month, the president had been more reticent about the tech, refusing to give a straight answer when asked for his opinion.
Flock has become a political landmine in the past year, as more Americans become alarmed over having a powerful, centralized surveillance network suddenly appear in their communities with little public oversight. Much attention has focused on how its over 120,000 cameras, which Flock frames as for strictly enforcing traffic laws and catching crime, are being abused.
A recent WaPo investigation found over fifty cases of police officers misusing Flock cameras and similar systems, including to stalk women. And Flock systems have either malfunctioned or been improperly used and caused innocent people — including a journalist — to be swarmed by armed cops. Such is the revulsion the company inspires that a grand jury in Ohio refused to indict a man who was caught destroying a Flock pole, despite surveillance footage clearly catching him in the act.
Trump’s endorsement is swooping in, in other words, while the company’s reputation is at its absolute nadir. Last month, its CEO Garrett Langley was forced to make a public apology after the WaPo investigation documented how its systems were frequently abused. Later, he insisted that his company was “not Big Brother,” before trying to reason with his critics by calling for a “compromise” between privacy and safety.
For Trump to be throwing his weight behind it now, without any perfunctory appeals to safety or concerns for Americans’ privacy, is quintessentially Trump, epitomizing how he has long abandoned any care for public opinion. (His full-throated support for data centers, which are even more unpopular than Flock, serve as a similar example.)
Even his Republican allies have taken a more cautious stance. Last month, Florida governor Ron DeSantis said the cameras are “out of control,” adding that he didn’t want “to have this become a surveillance state,” WaPo noted.
More on privacy: Police Investigator Accused of Making Over 4,000 Illegal Flock Camera Searches
The post Trump Throws His Weight Behind Flock Cameras appeared first on Futurism.
Read more of this story at Slashdot.
In case you thought the extreme backlash against automatic license plate readers (ALPRs) was excessive, a veteran police investigator was just arrested after conducting over 4,000 unauthorized searches using the software provided by Flock Safety.
According to New York’s NEWS10, Albany County Sheriff’s Office senior investigator Laurie Moore was arrested on Wednesday after conducting thousands of allegedly illegal searches on five different subjects.
One of the victims of Moore’s alleged ALPR abuse was reportedly her ex-girlfriend, whose license plate was reportedly queried in the system 3,000 separate times. Four additional vehicle plates were also searched hundreds of times each, bringing the grand total of unauthorized queries to over 4,000 across the last two years.
After being suspended from the police agency, Moore was charged with official misconduct and second-degree falsifying business records, both misdemeanors. Another unnamed employee of the Sheriff’s office was also placed on administrative leave, NEWS10 reported, though their fate is pending an internal investigation.
“She was an outstanding investigator, one of my absolute best employees,” Sheriff Craig Apple told local press. “That’s why it literally breaks my heart.”
Following Moore’s arrest, the agency said it revoked access to the Flock software for non-supervisory staff, and updated protocols to require an audit by Internal Affairs staff once every 30 days.
It’s just the latest in a deluge of cases involving police abuse of ALPR systems. Recently in Georgia, three officers and a staffer of the Savannah Police Department were arrested after an investigation uncovered similar misconduct, though details on the abuse haven’t been disclosed.
At the time of writing, the nonprofit Institute of Justice has logged 227 cases of ALPR abuse across the US.
More on Flock: People Taking Down Flock Towers Are Becoming Folk Heroes, With Communities Circling to Protect Them
The post Police Investigator Accused of Making Over 4,000 Illegal Flock Camera Searches appeared first on Futurism.
The surveillance state is trying out a new tactic: pocket watching.
404 Media reports on how secretive “predictive policing” units in the US Border Patrol are analyzing Americans’ financial activity and using it as a pretext to pull them over — even if they’re not suspected of a specific crime.
Predictive policing, the practice of identifying possible criminal activity by analyzing broader patterns in absence of direct evidence, is highly controversial. Delving into individual citizens’ financial data represents a significant escalation of these tactics.
According to the investigation, the Border Patrol units in question are called Predictive Intelligence Targeting Teams (PITT). Rather than conducting stops themselves, PITT units feed their findings to local law enforcement.
This was the case with a man driving across Montana, who was apprehended by local police after they received a tip-off from a PITT unit that analyzed his financial data. The man, Kyle William Olson, was stopped under the completely unrelated pretense, however, of an obstructed license plate. He was charged with a DUI and an intent to distribute narcotics. Olson had cannabis products like edibles in his car, which he got from a licensed cannabis farm in California where he worked, he told 404.
A memorandum of investigation document that Olson was provided and shared with 404 outlines how Border Patrol agent Matthew Phelps, who identifies himself as assigned to a PITT team in neighboring Spokane, Washington, observed “information contained within law enforcement-sensitive systems suggesting financial activity patterns commonly associated with illicit narcotics activity” when looking into Olson.
Experts criticized how PITT is disguising the dubious predictive techniques it’s using to apprehend Americans.
“The bottom line is genuine probable cause cannot be synthetically generated,” Jake Laperruque, deputy director of the Security and Surveillance Project at the Center For Democracy and Technology, told 404. Border Patrol, he added, is “using parallel construction to cloak the reason behind its car stops in secrecy.”
“If we can’t meaningfully review and evaluate these systems, we can’t trust them,” Laperruque said.
It remains unclear what mechanisms Border Patrol is using to keep tabs on Americans’ financial activity.
In a statement to 404, the agency said it “uses intelligence-informed analysis and planning to support national security operations, allocate resources, and help identify potential threats and bad actors to public safety,” but insisted it would not discuss its specific methods, sources of information, and other details, for “operational security reasons.”
More on privacy: Surveillance-Happy Police In Shambles After Trump Tariffs Makes Drones Absurdly Expensive
The post Federal Agents Are Secretly Monitoring Americans’ Financial Transactions So Cops Can Arrest People They Find Suspicious appeared first on Futurism.
Months after China’s Intermediate People’s Court ruled against a company using AI to liquidate an employee’s position, the country’s highest judicial body has come out with another whopper: a massive crackdown on AI deepfakes.
In a new ruling on Monday, China’s Supreme People’s Court — the top judicial organ in the country — issued new guidance on how judges in lower courts should handle civil issues arising from abuse of AI tools. As the South China Morning Post reports, the court’s opinion spans 24 articles, establishing official guidance on disputes around AI-generated video and audio, sexualized content, financial misconduct, digital privacy, and surveillance pricing.
One of the key platforms of the opinion revolves around deepfakes, which are lifelike images, video, and audio modeled after existing people — tech, as the ruling observes, that can easily be used to “bring the dead back to life.” This kind of content has exploded across the internet as sophisticated generative AI models become increasingly accessible to the general public, resulting in a staggering rise in scams, misinformation, and abusive content.
“In practice, many people express their strong concerns about AI deepfakes. AI face-swapping technology can indiscriminately harvest anyone’s likeness, making ordinary people potential targets of deepfakes,” Si Yanli, deputy head of the Supreme People’s Court’s research office, said in a statement. “AI voice cloning technology can also ‘steal’ voices at a very low cost, with results that can be convincingly realistic.”
Now, the SCMP notes that courts across the People’s Republic are being advised to view disputes around such content with an eye toward protecting individual identities, personal traits, and reputations.
Beyond abuses of AI, the new opinion also includes guidance on cases of doxing, or the practice of obtaining and spreading others’ personal information online.
“Doxing has emerged as a prominent form of cyberbullying, gravely violating victims’ privacy, disturbing their daily lives and eroding public security in cyberspace,” Si explained.
It’s worth noting that China uses a civil law system, so courts don’t follow the common-law principle of stare decisis. This means that, unlike the US, lower Chinese courts aren’t automatically bound by precedent set by other court rulings.
However, guidance issued by the Supreme People’s Court carries significant legal weight, with lower court bodies required to refer to such opinions in their own rulings — making the recent issuance a hefty rebuff against potential abuses of AI.
More on AI regulation: Dozens of Humanoid Robots Gather for a Protest, Marching and Waving Signs
The post Top Chinese Court Issues Sweeping Legal Crackdown on AI Deepfakes appeared first on Futurism.
Meta’s platforms are in a sorry state as grotesque AI slop continues to proliferate widely.
As a recent Futurism investigation found, many Facebook accounts have been flooding the platform with horrid videos depicting violent child abuse, and Meta appears to be woefully unprepared — and downright unwilling, judging by its ambiguously-phrased policies — to do anything meaningful about it.
At the same time, the Mark Zuckerberg-led company is finding plenty of new ways to creep out its user base. As part of the CEO’s lagging efforts to turn Meta into a viable competitor in the ongoing AI race, Meta launched a new AI assistant called Muse this week, and it’s already turning out to be deeply unsettling.
According to Meta’s marketing, the assistant is designed to “take the busywork off your plate,” such as booking tickets, making appointments, or filling out forms.
To actually reap any benefits, however, users have to hand over a staggering amount of trust — which, considering Meta’s catastrophic track record when it comes to protecting its users’ privacy, could easily turn out to be an absolutely terrible idea. While Meta claims the service runs on a secure “dedicated virtual machine” and that it will store login credentials in a place that cannot be accessed by the AI agent, the company has a lot of work to do to gain its users’ trust.
Perfectly illustrating the problem, as The Verge‘s Emma Roth found in her testing, Muse uncovered an “unnerving amount of information it autonomously gleaned about me.”
As part of her investigation, Roth linked both her Gmail and Amazon accounts to Muse and was immediately put off by how much personal info Muse pieced together right off the bat. It figured out the state where she was visiting her family, for instance, admitting it “pulled” data from the “shipping address on my Amazon order.”
It also listed off her “very specific interests” at a level of specificity that goes far beyond the “ad topics” section in the Instagram settings menu, which shows why certain advertisers target your account.
The experience left Roth “with an uneasy feeling about letting Muse sort through personal emails and handle my credit card.”
Meta’s creepy new AI chatbot perfectly illustrates how tech companies are continuing to push the boundaries of how much personal information we’re willing to give up just so we don’t have to complete mundane tasks — an extremely lopsided value proposition.
These same companies have repeatedly proven that they cannot be trusted with our private data. AI chatbots, which are turning out to be a surveillance-led cybersecurity nightmare in their own right, don’t instill much more confidence, either.
The post Meta Releases Uber-Creepy AI Chatbot as Its Platforms Crumble Under Grotesque Child Abuse appeared first on Futurism.
Apple says Live Rewind on Apple Watch cannot be used secretly because activation requires a deliberate double press of the Digital Crown, plays a chime, and shows a full-screen animation with a microphone indicator. The chime plays even when the watch is muted or headphones are connected. After activation, Live Rewind shows only a text […]
The post Apple Says Apple Watch Live Rewind Can’t Be Used Secretly appeared first on The Mac Observer.
Apple says Siri Recap can listen for conversations on Apple Watch, process the audio, and create a short high-level summary without keeping an audio recording for replay. The feature is opt-in, and users can control it by time, location, or a manual switch. Apple’s Audio Intelligence support document says raw audio is not recorded or […]
The post Siri Recap Privacy Explained: What Apple Does With Your Conversations appeared first on The Mac Observer.
At my gym, all the signs are handwritten and friendly. In one section, a sign asks members not to drop their weights because there are downstairs neighbors.1 In the bathrooms, reminders to flush and wash hands. But throughout the gym, signs say—more directly—no live streaming. Other signs ask people to exercise consideration when recording their routines for themselves or their coaches to ensure nobody else is in their sight lines. Privacy in the gym is paramount to ensure everyone is not self-conscious and is comfortable.
I was thinking of my gym when I heard about Live Rewind and Siri Recap in the just-announced Apple Watch Series 12 and Apple Watch Ultra 4. These are two of the four flagship Audio Intelligence features launching in beta later this year. Both of these tools record audio continuously. Live Rewind lets you double-press the Digital Crown to receive a transcription of the last 15 seconds of conversation. Siri Recap takes notes on what it hears throughout the day and provides Apple Intelligence-generated summaries.
Both features have, understandably, freaked people out, made them angry, or caused them to shrug in resignation. Surely Apple is exaggerating the privacy protections it explains on the same page it describes the features! And if Apple is all in on recording us all the time without our consent or knowledge whenever one of its devices is nearby, the privacy game is over, isn’t it!? Everything becomes fodder for the machine; private conversations are immutable; the future is surveillance!
I’m not one for extremes, and I’ve studied Apple’s privacy and security provisions and mistakes for nearly 30 years. But I have concerns about these features from social, political, safety, and legal standpoints. Let me tackle the privacy and security first.
When Apple announced Private Cloud Compute in 2024, it was interesting, but it felt academic because it didn’t then release a Siri worth worrying about. In 2026, the PCC expansion with Siri AI is much more important, as Siri may send huge amounts of our personal information to the cloud to provide better answers derived from our data.2
After studying Apple’s details on PCC, I felt reassured. This is partly because one of Apple’s five pillars of PCC integrity is transparency. This is unlike many previous claims by Apple that it would be transparent. Here, they’ve stated exactly what will be available to researchers, and one researcher already found a flaw and was awarded the highest Apple bounty—$150,000. The other pillars include end-to-end encryption, barring access by Apple or other parties through cryptographic design, and no way for an external (or internal!) party to determine which PCC server is processing an individual’s data.
Apple’s promises with Audio Intelligence are somewhat different, though I find them a valid approach to minimizing invasions of one’s privacy and an attempt to reduce violations of other people’s personal sphere. Here are the fundamentals for Live Rewind and Siri Recap:
Audio isn’t recorded, nor is it stored, nor is a transcript made. As it is captured, the audio is not accessible to you, watchOS or iOS, apps, or Apple. Apple achieves this by using the Secure Exclave to process audio; after processing, the audio is immediately discarded.4
All of the audio processing happens on your Apple Watch and iPhone. That prevents leakage. If you engage Siri, however, some portion of the Live Rewind text could be sent to PCC under the same security protections. (Apple says “will.”)
Siri Recap uses PCC. The process is complicated, exposing as little information as possible within the already secured PCC environment.
iCloud syncing is end-to-end encrypted. You don’t need to enable Advanced Data Protection (ADP).
Siri Recap deletes summaries after seven days; Live Rewind’s 15-second text snippets are deleted within 30 seconds after the Apple Watch display dims. You can save Siri Recaps or export them. With Live Rewind, they are retained within the Siri app if you choose to save them or ask Siri about the text.
From a design standpoint, these measures all feel well thought out to prevent all of the egregious use cases one might imagine: Unwanted recordings! Apple analyzing our conversations to sell us stuff (or sell to third parties)! Leakage that would allow friends or strangers to access summaries, snippets, or conversations, including via internet cloud storage!
And yet. I find the above fits into the meme of “My ‘An explanation of how the Apple Watch recording what you are saying is kept private and anonymous’ T-shirt has people asking a lot of questions already answered by my shirt.”
When I’m out in public in the United States, courts have found that in most situations, I have little right to privacy. But then you can shave that down. If a neighbor points a camera from their house at my unfenced front lawn, visible from the street?Generally legal, with some exceptions related to harassment or intimidation. But if a camera points into my house through a window or into a fenced area of my property, most jurisdictions find that “not cool” (a legal term).
The audio side of this is very different. My state of Washington, for instance, is an “all-party consent” state; some states are “one-party consent” or have different rules. “One party” means anyone in the conversation (on a phone, in person, via videoconferencing, etc.) can consent, so they can record audio or have a device or system do it for them. In Washington and other all-party states, every person being recorded must consent, or it is illegal and actionable. (Recording across state lines or internationally becomes complicated.)
The company says it’s not “recording” or “storing” audio. But it is capturing it. Shockingly, I’m not a lawyer, so I can’t provide a legal analysis. (Presumably many privacy-focused organizations are doing deep reads of state-level statutes.) But the gut check is what I’m really going for, anyway: do these two features have the feeling of being recorded without one’s permission, even if Apple is trying to mitigate both the technical reality and the sense of it?
On the technical side, it appears that you have to enable each feature in the Watch app on the paired iPhone. Live Rewind requires enabling Siri: Live Rewind: Double Click Digital Crown. You can schedule Siri Recap by time, trigger it or turn it off using a location, or enable or disable it manually, such as through Control Center.
So let’s do the vibe check.
Live Rewind appears to be buffering audio at all times in anticipation of needing to transcribe the last 15 seconds. Apple says Live Rewind is designed to alert those around you that you’re replaying audio: when you double-click the Digital Crown, a chime sounds (ignoring any of your settings) as the Apple Watch’s entire screen is taken over with an animation that includes a microphone indicator.
However, unless Live Rewind is also a literal Time Machine, this doesn’t feel right. People are being warned 15 seconds after their audio is captured. (If Apple doesn’t mean that, it needs to clarify.) Further, it assumes that people will be keyed in to someone’s smartwatch making sounds and flashing images they should be paying attention to.
There’s also no “stop recording” option: if you’re being recorded by Live Rewind, you as an unrelated party—a waiter, a friend, a stranger, someone nearby, a celebrity—have no way to say “no.”
Siri Recap ostensibly is capturing audio whenever it’s active in any of the circumstances above. There’s no indication to other people that it’s running at all, and no one except the user can opt out.
Apple’s mitigations here include that no speaker is attributed, though Siri Recap identifies and anonymously labels different voices.
With Siri Recap, an Apple Watch automatically discards “potentially harmful content.” Apple defines this as content that “promotes self-harm or hateful speech, and sensitive information, such as financial data, government-assigned identifiers, authentication data, and certain personal identifiers.” I’m not sure this list is a strong argument for Siri Recap.
When using a public or private space, there should be a compelling case, even with the protections enumerated. There are already many kinds of assistive technology used by people with many kinds of disabilities to let them function better in a world often hostile to people with no or low vision or hearing.
However, the assistive case isn’t what defines Live Rewind, a useful feature of convenience that has massive privacy implications despite the way data is captured and stored. And it doesn’t help with Siri Recap at all. Siri Recap is just “life recording” without the label attached, since no audio or video is retained, and only 0.01% of a day’s conversation will likely be summarized.
These features feel like they were designed in a lab and passed through to production without enough people wondering, “Will this provoke a tremendous backlash because of the gaps in privacy that are relatively obvious?”
The best way I can conclude my analysis is by quoting Apple:
Be mindful of those around you. While audio is not recorded and stored, and no speakers are attributed, consider those around you where conversations might be private or sensitive.
Given that Apple has to warn of this, it’s the surest sign the features shouldn’t be released in this form. Apple repeats the form of the warning I see regularly at my gym… but apparently it isn’t heeding it.
Read more of this story at Slashdot.
Login