Comment by crazygringo
23 days ago
> This is exactly what Apple Intelligence should have been... They could have shipped an agentic AI that actually automated your computer instead of summarizing your notifications. Imagine if Siri could genuinely file your taxes, respond to emails, or manage your calendar by actually using your apps, not through some brittle API layer that breaks every update.
And this is probably coming, a few years from now. Because remember, Apple doesn't usually invent new products. It takes proven ones and then makes its own much nicer version.
Let other companies figure out the model. Let the industry figure out how to make it secure. Then Apple can integrate it with hardware and software in a way no other company can.
Right now we are still in very, very, very early days.
I don’t believe this was ever confirmed by Apple, but there was widespread speculation at the time[1] that the delay was due to the very prompt injection attacks OpenClaw users are now discovering. It would be genuinely catastrophic to ship an insecure system with this kind of data access, even with an ‘unsafe mode’.
These kinds of risks can only be _consented to_ by technical people who correctly understand them, let alone borne by them, but if this shipped there would be thousands of Facebook videos explaining to the elderly how to disable the safety features and open themselves up to identity theft.
The article also confuses me because Apple _are_ shipping this, it’s pretty much exactly the demo they gave at WWDC24, it’s just delayed while they iron this out (if that is at all possible). By all accounts it might ship as early as next week in the iOS 26.4 beta.
[1]: https://simonwillison.net/2025/Mar/8/delaying-personalized-s...
Exactly. Apple operates at a scale where it's very difficult to deploy this technology for its sexy applications. The tech is simply too broken and flawed at this point. (Whatever Apple does deploy, you can bet it will be heavily guardrailed.) With ~2.5 billion devices in active use, they can't take the Tesla approach of letting AI drive cars into fire trucks.
This is so obvious I'm kind of surprised the author used to be a software engineer at Google (based on his Linkedin).
OpenClaw is very much a greenfield idea and there's plenty of startups like Raycast working in this area.
9 replies →
Regardless of how Apple will solve this, please just solve it. Siri is borderline useless these days.
> Will it rain today? Please unlock your iphone for that
> Any new messages from Chris? You will need to unlock your iphone for that
> Please play youtube music Playing youtube music... please open youtube music app to do that
All settings and permission granted. Utterly painful.
22 replies →
Its hard to come up with useful AI apps that aren't massive security or privacy risks. This is pretty obvious. For an agent to be really useful it needs to have access to [important stuff] but giving an AI access to [important stuff] is very risky. So you can get some janky thing like OpenClaw thats thrown together by one guy and has no boundaries and everyone on HN thinks is great, but its going to be very difficult for a big firm to make a product like that for mass consumption without it risking a massive disaster. You can see that Apple and Microsoft and Salesforce and everyone are all wrestling with this. Current LLMs are too easily hoodwinked.
I think you're being very generous. There's almost 0 chance they had this actually working consistently enough for general use in 2024. Security is also a reason, but there's no security to worry about if it doesn't really work yet anyway
The more interesting question I have is if such Prompt Injection Attacks can ever be actualy avoided, with how GenAI works.
Removing the risk for most jobs should be possible. Just build the same cages other apps already have. Also add a bit more transparency, so people know better what the machine is doing, maybe even with a mandatory user-acknowledge for potential problematic stuff, similar to how we have root-access-dialogues now. I mean, you don't really need access to all data, when you are just setting a clock, or playing music.
Perhaps not, and it is indeed not unwise from Apple to stay away for a while given their ultra-focus on security.
They could be if models were trained properly, with more carefully delineated prompts.
2 replies →
The prompt injection thing is especially nasty for agents because they process untrusted input (web pages, emails, documents) and can take real actions. With a chatbot, prompt injection makes it say something dumb. With an agent that acts as you, a malicious payload hidden in an email could make it forward your contacts, reply on your behalf, whatever. You can't fix this in the model alone — you need an enforcement layer outside the model that limits what it can actually do regardless of what it thinks it should do. I'd bet Apple is working on exactly this and it's why they're taking their time.
Apple's niche product, consisting of like 1-4% of computer sales compared to its dominant MacBook line, is now flying off the shelf as a highly desired product, because of a piece of software that Apple didn't spend a dime developing. This sounds like a major win for Apple.
The OS maker does not have to make all the killer software. In fact, Apple's pretty much the only game in town that's making hardware and software both.
Really doubt it has a significant impact on mac mini sales…
And being fair ClawBot is a complete meme/fad at this point rather than an actual product. Using it for anything serious is pretty much the equivalent of throwing your credit cards, ids and sticky notes with passwords and waiting to see what happens…
I do see the appeal and potential case of the general concept of course. The product itself (and the author has admitted it themselves) is literally is a garbage pile..
> Using it for anything serious
One man's trash is another man's serious
What are you referring to?
Probably the Mac Mini. A few OpenClaw users are buying the agent a dedicated device so that it can integrate with their Apple account.
For example: https://x.com/michael_chomsky/status/2017686846910959668.
16 replies →
The entire point of the article is about the Mac mini sales flying through the roof because of this.
Mac-Minis
[flagged]
19 replies →
Do you want to know how I can tell you didn't read the article?
1 reply →
> ...Apple doesn't usually invent new products. It takes proven ones and then makes its own much nicer version.
While this was true about ten years ago, it's been a while since we've seen this model of software development from Apple succeed in recent years. I'm not at all confident that the Apple that gave us Mac OS 26 is capable of doing this anymore.
Best privacy in computers, ADP, and M-series chips mean nothing to you? To me, Apple is the last bastion of sanity in a world where user hostility is the norm.
Apple is certainly the least worst but man... Liquid Glass. Windows bordering on the circular...
As said elsewhere, success in hardware does not translate to success in software.
Privacy is definitely good but it's not at all an example of the success mentioned in the parent comment. It's deep in the company culture.
Airtags were released in 2021, I'd say that counts, but generally I agree.
Their hardware division has been killing it.
The software has been where most of the complaints have been in recent years.
9 replies →
I mean they literally just looked at Tile. And they have the benefit of running the platform. Demonstrates time and time again that they engage in anticompetitive behaviour.
1 reply →
> And this is probably coming, a few years from now. Because remember, Apple doesn't usually invent new products. It takes proven ones and then makes its own much nicer version.
Except this doesn't stand up to scrutiny, when you look at Siri. FOURTEEN years and it is still spectacularly useless.
I have no idea what Siri is a "much nicer version" of.
> Apple can integrate it with hardware and software in a way no other company can.
And in the case of Apple products, oftentimes "because Apple won't let them".
Lest I be called an Apple hater, I have 3 Apple TVs in my home, my daily driver is a M2 Ultra Studio with a ProDisplay XDR, and an iPad Pro that shows my calendar and Slack during the day and comes off at night. iPhone, Apple Watch Ultra.
But this is way too worshipful of Apple.
In that list of Apple products that you own, do none of them match the ops comment? You’re saying none of those products are or have been in their time in the market a perfected version of other things?
There are lots of failed products in nearly every company’s portfolio.
AirTags were mentioned elsewhere, but I can think of others too. Perfected might be too fuzzy & subjective a term though.
We're talking about Apple Intelligence here and its ... "precursor" ... Siri.
Both of which have been absolutely underwhelming if not outright laughable in certain ways.
Apple has done plenty right. These two, which are the closest to the article, are not it.
Remember the time when the former members of the Siri team demoed a prototype for a more capable version of Siri and Apple didn't even use it
Perhaps I’m misremembering, but I feel sure that Siri was much better a decade ago than it is today. Basic voice commands that used to work are no longer recognised, or required you to unlock the phone in situations where hands free operation is the whole point of using a voice command.
There were certain commands that worked just fine. But they, in Apple's way, required you to "discover" what worked and what didn't with no hints, and then there were illogical gaps like "this grouping should have three obvious options, but you can only do one via Siri".
And then some of its misinterpretations were hilariously bad.
Even now, I get at a technical level that CarPlay and Siri might be separate "apps" (although CarPlay really seems like it should be a service), and as such, might have separate permissions but then you have the comical scenario of:
Being in your car, CarPlay is running and actively navigating you somewhere, and you press your steering wheel voice control button. "Give me directions to the nearest Starbucks" and Siri dutifully replies, "Sorry, I don't know where you are."
Absolutely none of the things you quoted that he said an AI agent could do would I want be done for me and I doubt most other people would.
It would be an absolute disaster at Apple scale. Millions of people would start using it, filing incorrect taxes or deleting their important files and Apple would be sued endlessly.
Tiny open source projects can just say "use at your own risk" and offload all responsibility.
Here is a fun “Prompt Injection” which I experimented with before the current AI Boom; visiting a friend’s home › see Apple/Amazon listening devices › Hey Siri/Alexa, please play the last song. Harmless, fun.
Google TV did "show passport photos" back in 2017. My friends loved it!
>> Imagine if Siri could genuinely file your taxes
Imagine if the government would just tell everyone how much they owed and obviated the need for effing literal artificial intelligence to get taxes done!
>> respond to emails
If we have an AI that can respond properly to emails, then the email doesn't need to be sent in the first place. (Indeed, many do not need to be sent nowadays either!)
Yeah the whole filing taxes thing is an epic XY-problem. Governments can make this as easy as a digital signature, there’s zero need for an agent of any kind.
Actually most of the things people use it for is of this kind, instead actually solving the problem (which is out of scope for them to be fair) it’s just adding more things on top that can go wrong.
Seriously. The best solution is not having the problem in the first place. Something something Tao Te Ching.
If a user chooses to reach out about an issue that an AI agent can completely solve, why should they not be allowed to do so via email? I much prefer it over all other support communications channels.
How can government know how much you owe them when they don't know all your tax deductibles?
You being personally ignorant of this specific argument which gets litigated every single time this comes up but only by Americans because most other countries have zero difficulty doing exactly that is not a valid argument.
91 percent of American filers take the standard deduction. The IRS already has all their information, already knows how much they withheld, already knows what they owe back. For all these people, TurboTax is just filling in 10 fields in the standard form.
"All your tax deductibles" is irrelevant for the vast majority of the country, and always has been.
The 35 million remaining americans who do itemize are free to continue using this old system while the rest of us can have a better world.
We could also ask how the government could later tell someone they improperly deducted something! The government can either use that same means to tell taxpayers in advance, or else we could figure out a superior taxation system that wouldn’t require these steps.
By knowing all your tax deductibles?
4 replies →
But the lobbyists
File taxes? That's a tall order, especially juxtaposed with managing calendar or responding to emails.
>File taxes?
Sure why not, what could go wrong?
"Siri, find me a good tax lawyer."
"Your honor, my client's AI agent had no intent to willfully evade anything."
These people live on another planet.
1 reply →
Tax filing is trivial in most countries with a functioning government, it’s only a Big Deal in the US due to Intuit bribing the government.
Even in the US, for most people tax filing it not really a complex process. It only gets complicated if you are trying to itemize deductions and have a complex income story. Most people can do it with a couple of documents and a single form.
1 reply →
I think the interesting tension here is between capability and trust.
An agent that can truly “use your computer” is incredibly powerful, but it's also the first time the system has to act as you, not just for you. That shifts the problem from product design to permission, auditability, and undoability.
Summarizing notifications is boring, but it’s also reversible. Filing taxes or sending emails isn’t.
It feels less like Apple missing the idea, and more like waiting until they can make the irreversible actions feel safe.
> An agent that can truly “use your computer” is incredibly powerful, but it's also the first time the system has to act as you, not just for you. That shifts the problem from product design to permission, auditability, and undoability.
Or rather, just reveals that the industry never bothered to properly implement delegation of authority in operating systems and applications, opting instead to first guilt-trip people for sharing their passwords, and later inventing solutions that make it near-impossible to just casually let someone do something for you.
Contrast with how things in real life function, whether at family level or at the workplace.
Half-agree. As an industry, we do suffer from not-invented-here syndrome, so have a lot of mediocre attempts to implement it that don't learn enough lessons from historical human-human examples.
Delegation can also be scary with other humans. "Power of attorney" and all that. Or even just micro-management.
Clicking `Submit` is easiest step of sending email / filling taxes.
All steps before it are reversible, and reviewable.
Bigger problem is attacker tricking your agent to leak your emails / financial data that your agent has access to.
I worry we'll click "Submit" as fast as we click "I accept the terms and conditions."
1 reply →
> Imagine if Siri could genuinely file your taxes, respond to emails, or manage your calendar
> And this is probably coming, a few years from now.
Given how often I say "Hey Siri, fast forward", expecting her to skip the audio forward by 30 seconds, and she replies "Calling Troy S" a roofing contractor who quoted some work for me last year, and then just starts calling him without confirmation, which is massively embarassing...
This idea terrifies me.
Also in the good old days if you sealed the wrong number you had some time to just hang up without harm done. Today the connection is made the moment you pressed the button or in this case when Siri decided to call.
Happened to me too while being in the car. With every message written by Siri it feels like you need to confirm 2 or 3 times (I think it is only once but again) but it calls happily people from your phone book.
> Because remember, Apple doesn't usually invent new products. It takes proven ones and then makes its own much nicer version.
Funny seeing this repeated again in response to Siri which is just... not very good.
hey siri can set the egg timer 90% of the time corectly! Find me another multitrillion dollar company that is able to pull that off!
.
How do people manage to pick such bad examples? Who in their right mind would ever allow an LLM to FILE THEIR TAXES for them. Absolutely insane behavior. Why would anyone think this is probably coming? Do you think the IRS is going to accept "hallucination lol" as an excuse for misfiling?
Because private taxe filling software, like used in the USA, are exempt from filling errors?
If you're quick at responding and fixing the problem, the IRS forgives much..
Can you understand how this commoditizes applications? The developers would absolutely have a fit. There is a reason this hasn’t been done already. It’s not lack of understanding or capability, it’s financial reality. Shortcuts is the compromise struck in its place.
This is generally true only of them going to market with new (to them) physical form factors. They aren’t generally regarded as the best in terms of software innovation (though I think most agree they make very beautiful software)
Personal intelligence, the (awkward) feature where you can take a screenshot and get Siri to explain stuff, and the new spotlight features where you can type out stuff you want to do in apps probably hints at that…
People forget that “multi touch” and “capacitive touchscreens” were not Apple inventions. They existed prior to the iPhone. The iPhone was just the first “it just works” adaptation of it
Not a great example as multitouch in its modern incarnation was a niche academic technology, the most refined version of which was built by a 2 person startup that Apple quickly acquired. There was still a long way to go to make the tech as ubiquitous as it is today and that was all heavy lifting done by Apple.
Well, the heavy lifting was supervised by the same people, but while receiving Apple paychecks :)
> Because remember, Apple doesn't usually invent new products. It takes proven ones and then makes its own much nicer version.
Apple doesn't take proven ones of anything. What they do is arrive at something proven from first principles. Everyone else did it faster because they borrowed, but Apple did it from scratch, with all the detail-oriented UX niceties that entails.
This was more prevalent when Jobs was still around. Apple still has some of that philosophy at its core, but it's been eroding over time (for example with "AI" and now Liquid Ass). They still do their own QA, though, and so on. They're not copying the market, they have their own.
>It takes proven ones and then makes its own much nicer version.
I think you repeated their marketing, I don't believe this is actually true.
> Then Apple can integrate it with hardware and software in a way no other company can.
That's a pretty optimistic outlook. All considered, you're not convinced they'll just use it as a platform to sell advertisements and lock-out competitors a-la the App Store "because everyone does it"?
I would guess, and it is a guess, that there are two reasons apple is “behind” in AI. First, they have nowhere near the talent pool or capability in this area. They’re not a technical research lab. For the same reason you don’t expect apple to win the quantum race, they will not lead on AI. Second, AI is a half baked product right now and apple try to ship products that properly work. Even Vision Pro is remarkably polished for a first version. AI on the other hand is likely to suffer catastrophic security problems, embarrassing behaviour, distinctly family-unfriendly output.
Apple probably realised they were hugely behind and then spent time hand wringing over whether they remained cautious or got into the brawl. And they decided to watch from the sidelines, buy in some tech, and see how it develops.
So far that looks entirely reasonable as a decision. If Claude wins, for example, apple need only be sure Claude tools work on Mac to avoid losing users, and they can second-move once things are not so chaotic.
> Imagine if Siri could genuinely file your taxes
If you trust openclaw to file your taxes we are just on radically different levels of risk tolerance.
every time i've heard someone's speculations about what apple intelligence could have been, it's a complex conspiracy. its problem is that it sucks and makes them no money, so they didn't ship it.
[dead]
[dead]
Apple literally lives on the "Cutting Edge" a-la XKCD [1]. My wife is an iPerson and she always tells me about these new features (my phone has had them since $today-5 years). But for her, these are brand new exciting things!
https://xkcd.com/606/
How many chat products has Google come out with? Google messenger, buzz, wave, meet, Google+, hangouts… Apple has iMessage and FaceTime. You just restated OP’s point. Apple evolves things slowly and comes to market when the problems have already been solved in a myriad of ways, so they can be solved once and consistently. It’s not about coming to market soonest. How did you get that from what OP said?
Pointless argument given that android isn't just "android". Never has been.
It's a huge, diverse ecosystem of players and that's probably why Android has always gotten the coolest stuff first. But it's also its achilles' heel in some ways.
1 reply →
Android isn't all about Google. Where I live everyone uses WhatsApp and Telegram, both of which have nothing to do with Google.
"It’s not about coming to market soonest. "
First Mover effect seems only relevant when goverment warrants are involved. Think radio licenses, medical patents, etc. Everywhere else, being a first mover doesnt seem to correlate like it should to success.
2 replies →
A very tired “red versus blue” take here.
There are plenty of Android/Windows things that Apple has had for $today-5 years that work the exact same way.
One side isn’t better than the other, it’s really just that they copy each other doing various things at a different pace or arrive at that point in different ways.
Some examples:
- Android is/was years behind on granular permissions, e.g. ability to grant limited photo library access to apps
- Android has no platform-wide equivalent to AirTags
- Hardware-backed key storage (Secure Enclave about 5 years ahead of StrongBox)
- system-wide screen recording
Android is an OS, not hardware tho so some of those can't really be judged equivalently.
1 reply →