Control Mac with Voice - Trying

Kinda been almost a year since your first post to this forum announcing that you were making such a thing yourself. Did you give up and are just promoting a competitor, or is just a renamed attempt? It's under the same domain so I'm guessing it's the same old thing that was coming out in June last year or whatever.

Sorry, but although you don't remember me, which is kind of a good thing, I do remember you very well... Hope that thing was not vibe coded, as that would make it significantly worse considering how much privilege over people's computers it requires to accomplish its intended purpose, and I still find it quite weird for you to not have a single demo video published.

Applications distributed on the App Store are reviewed, notarized applications are just automatically verified mostly for malware, so security issues, unintended behavior, and privacy violations aren't guaranteed to be checked for and much less caught at all. The Transparency, Consent, and Control / Gatekeeper prompt only states that the code was verified for the presence of malware, so if anyone decides to install this, beware of what you choose to reply to any security prompts.

You definitely don't remember me, otherwise you'd be well aware that I know exactly what I'm talking about, and to drive the point home, I'll just ask my local ChatGPT:

jps@alpha ~ % mlx_lm.chat --max-tokens=1048756 --model models/gpt-oss-120b 
/Users/jps/Library/Python/3.9/lib/python/site-packages/urllib3/__init__.py:35: NotOpenSSLWarning: urllib3 v2 only supports OpenSSL 1.1.1+, currently the 'ssl' module is compiled with 'LibreSSL 2.8.3'. See: https://github.com/urllib3/urllib3/issues/3020
  warnings.warn(
[INFO] Starting chat session with models/gpt-oss-120b.
The command list:
- 'q' to exit
- 'r' to reset the chat
- 'h' to display these commands
>> Be terse and direct to the point answering my following questions.
<|channel|>analysis<|message|>The user says: "Be terse and direct to the point answering my following questions." So they want terse, direct answers. We need to wait for their questions. So we should respond acknowledging and ask for the questions? Or we can say "Sure, ask your questions." The user hasn't asked any question yet. So we can respond briefly: "Sure, go ahead." That's appropriate.<|end|><|start|>assistant<|channel|>final<|message|>Sure—go ahead with your questions.
>> Some guy on the Internet claims that Apple's notarization process ensures that applications are entirely safe of security and privacy issues. Is this true?
<|channel|>analysis<|message|>The user asks: "Some guy on the Internet claims that Apple's notarization process ensures that applications are entirely safe of security and privacy issues. Is this true?" They want a terse, direct answer. The user previously asked to be terse and direct. So we should answer concisely, but also provide enough nuance. The answer: No, notarization is not a guarantee of safety; it only checks for known malware, code signing, and some basic security checks, but does not guarantee privacy, no vulnerabilities, no malicious behavior, etc. Provide brief points: Notarization is a gatekeeper, but not a full security audit; it doesn't guarantee no zero-days, no data collection, no privacy issues; it only ensures the app is signed, not tampered, and passes basic checks. Also note that notarization can be bypassed, and malicious apps can still be notarized if they pass checks. So answer: No, it's not true. Provide short explanation.<|end|><|start|>assistant<|channel|>final<|message|>No. Apple’s notarization is a gate‑keeper, not a guarantee of safety.

* **What notarization does** – Apple scans the binary for known malware, checks that it’s properly code‑signed, and verifies it can run on macOS without violating basic system policies. If it passes, Apple issues a notarization ticket that lets Gatekeeper allow the app to run without a warning.

* **What notarization doesn’t do** –  
  • It doesn’t audit the app’s source code or logic, so hidden bugs, zero‑day vulnerabilities, or deliberately malicious behavior can still be present.  
  • It doesn’t guarantee that the app respects user privacy; an app can collect data, track you, or exfiltrate information and still be notarized as long as it isn’t flagged as malware.  
  • It isn’t a formal security review; the checks are automated and limited to known signatures and heuristics.

In short: notarization reduces the risk of obvious malware, but it **does not make an app “entirely safe”** from security flaws or privacy‑invasive practices. Always evaluate apps on their own merits, permissions, and reputation, even if they’re notarized.
>> Alright, thanks! Bye!
<|channel|>analysis<|message|>User says "Alright, thanks! Bye!" They are ending conversation. We should respond politely, maybe a brief farewell. Keep it short.<|end|><|start|>assistant<|channel|>final<|message|>You’re welcome—take care!
>> q

That said, this application is fully code-signed with a valid Developer ID certificate and successfully notarized by Apple. It passes macOS Gatekeeper verification without warnings.

Which means absolutely nothing, as I explained above.

If you have any specific technical concerns about its behavior, I would be happy to address them directly. General statements about notarization limitations do not imply that this particular application is unsafe.

Of course not, but implying that notarization addresses any of the security concerns that I raised is shady at least, since you are clearly deceiving people by providing a false sense of security. Code-signing is required for the notarization process, so it doesn't make sense to imply that it's something different. Either of these two shady practices alone are enough for me to recommend people to stay clear of anything that you build, together they form a huge red flag that nobody should ignore, and that's without even talking about the astroturfing tone in which you wrote the original post.

I know you're also developing a macOS app, and I'd like to offer you a collaboration. I can send you an app, and you can check it for all malware, which you can only check. If you're convinced the app is completely valid, I'd like you to write in the comments that it's absolutely valid. I'd like to suggest we continue working together. What do you think about this?

I'm not working on that anymore, as I stepped out of that job back in November, after a year of freelance work, to pursue my own business, plus and although I'm perfectly comfortable to do reverse-engineering, I'm not particularly fond of it, it's just another tool in my problem-solving toolbox, otherwise I'd have followed the cybersecurity route a long time ago. These days I call my own shots and that's exactly how I want it to be, as I have my own ideas of how AI should be built and sold, the ability to bring those ideas to life and verify their feasibility, and the resources to make it all happen, so I'm a fully self-sufficient entrepreneur.

I believe it’s important to avoid making accusations without concrete technical evidence. Broad statements suggesting malicious intent can be misleading, especially in a public discussion.

Feel free to quote any of my earlier comments making accusations. At most I pointed at red flags, I never accused you of making malware, but I'm fulfilling my social obligation to warn people about the dangers of running this kind of software especially given how you chose to announce it here.

I also want to clarify mine: I have not claimed that notarization guarantees absolute security, nor have I presented it as such. I stated it as a factual part of the distribution process.

OK then, so what did you intend to say with the quote below?

I also want to clarify that you have nothing to doubt because Apple does not allow you to distribute applications just like that. That is, what does it have to do with distributing applications? I have to get it approved by Apple and sign all the documents. Only after that can I distribute it when Apple is sure that it is absolutely safe. You can check this and ask ChatGPT, so you have nothing to doubt.😊

Those are your words, not mine, I did not make any accusations myself, only refuted your clearly misleading statements that to me came across as implying that notarized applications faced the same kind of scrutiny from Apple as App Store applications, while also claiming that App Store applications themselves are guaranteed to be safe.

You’re free to inspect the binary, review the libraries, analyze network behavior, or test any aspect you consider relevant. If you identify weaknesses or areas that could be improved, I’m open to concrete, constructive feedback.

How much are you paying for that service? If it's sufficiently appealing to me, I will put in the effort, to add extra funding to my own business, but you'll have to pay 20% of the total amount in advance, otherwise I won't spend any time providing this kind of service for free. There are plenty of cybersecurity companies out there employing people who actually enjoy doing that kind of job for a living, which as I said earlier is not my case, but am willing to make the sacrifice for the right offer.

If you truly remember me, you should remember that I actually despise agentic AI, it just makes no sense to me, not only due to the privacy issues, but also because it has a huge attack surface and is prone to hallucinating. Consulting AI to get the initial leads to start a research and learn new things, quickly searching content with Retrieval-Augmented Generation, enlisting its assistance to review one's work, and tirelessly flagging outliers in huge streams of information for human review, are all proper use cases for the technology, but switching the roles by allowing machines to get sloppy while we review and take responsibility for their actions is not the way to go, and this hype can't die soon enough.

Currently people seem to be focused not in making AI that empowers human beings by taking over the tasks where computers truly excel but rather in making AI that attempts to take over tasks where humans have always been superior to machines, while doing a worse job and relegating humans to the validating positions where machines have always excelled. This is not advancing civilization, quite the opposite, it's dumbing us down collectively, even all the senior engineers who chose to drink the Kool-Aid. I don't mind this personally, because once this bubble inevitably bursts, I'll be left standing largely unaffected with a sharp brain and at the top of my game, but it saddens me to watch how ridiculously stupid this timeline is getting.

In these cases, the AI is not replacing human intelligence — it is reducing mechanical overhead.

All those examples can be easily tackled by deterministic and highly efficient algorithmic solutions, there's no need to use extremely inefficient and error prone heuristics based on machine learning for any of them. Agentic AI is a solution looking for a problem, a gimmick to fool the ignorant, not an actual tool.

Hi Sergiy,

You've mentioned that you welcome feedback on the app's security and privacy, and I'm always glad to see devs being so open to reviews of their hard work. Have you made your source code available for viewing directly? I've checked GitHub and GitLab and haven't been able to find it, so I'd be grateful if you could provide a link to the project.

Thanks in advance for your time and transparency.

The problem here is not the AI itself, the problem is with its agentic use which, as I said, is making computers act all sloppy and highly inefficiently in lots of cases where they could just be acting like highly accurate deterministic systems, and the worst part of it is that us, humans, are being relegated to reviewing that slop, which is what machines are actually good and have historically been used for. The proper way to use AI is to make it do the review where that makes sense, not to do our jobs with a worse performance than a professional would.

While to you it might sound like agentic AI is a cheap way to get professional work done, that is only true if you ignore the fact that 90% of the investment into that technology is actual lost, as well as that so far nothing of real value has actually been made with it, so the only reason why you can even think about avoiding human workers is because investors are promoting the wrong kind of business that will eventually blow up in their faces, and you're tagging along for the ride. Not only that but a whole generation of software engineers are getting insufficient training because of this crazy social experiment. Finally, I think that the consequence of flooding a market with low quality products will be the sinking of that low-end market, not an increase in revenue for the low-end producers, because the whole thing will just turn into a race to the bottom for the cheapskates.

This guy is either teasing people to review his code for free or taking advantage of the fact that nobody will actually bother to do it without compensation to back his safety claims. The point here is that the way he presented himself, pretending to be a member of the community promoting his product, a marketing strategy known as astroturfing, is really shady, and if you couple that with his earlier comments essentially trying to convince me of all the people that his product is safe, raises lots of red flags that people should definitely ignore at the very least until the product is certified by a reputable cybersecurity company.

Nobody will take the bait to review your code for free, forget about that, it's not going to happen, and your requests to have people do it won't add any credibility to your safety story. I mean I can just ask a simple question that you will most likely reply negatively to, which is whether you are running the models locally, because if you aren't, then you are already exfiltrating data out of people's computers, to a drunken robot that can easily hallucinate and whoever knows what happens after. One of the problems with agentic AI is that it's unsafe by design, AI itself is based on heuristics, not determinism, and the sheer complex of large language models means that no human being can claim safety with any level of certainty, so trusting one's digital life to those models is reckless to say the least.

Be aware of any software that claims to run locally, if it still requires Internet access for any reason. That too can be a security risk.
@João Santos, I know that you already know this, just sharing with the general public. 😇

This community does not allow direct product promotion posts, which is why the post was written in the format it was. I didn’t have another way to introduce the project here while respecting the group rules.

So you chose to pretend to not be breaking the rules? This would make good comedy if it wasn't so sad... I'm literally facepalming in disbelief right now, as it tells a lot about your personality...

Hi Sergiy,

I wanted to take a moment to publicly offer information about our guidelines surrounding developer engagement on our platform.

We welcome and encourage developers to post to the forum to introduce their app to our community, explain what it does, solicit feedback, and answer any user questions. An example of this type of post might look like:

Dear AppleVis Community,

My name is Michael, and I have just released a weather app with full VoiceOver support. Below is a list of the app's features:

[Insert feature list.]

If you have any questions, please do let me know! I welcome any and all feedback on how the app can be improved, both in terms of features and accessibility. Here is a link to the app store. [Inserts App Store link.]

The one limitation we place on developer posts specifically is that we ask that developers please not add their own apps to our App Directories. An App Entry is akin to a product review, and it is essential that reviews on our platform be authentic and represent the lived experience of product end-users.

For now, I can send you the application itself so you can test or check it. If after testing you still feel it’s necessary to review the source code, you can send me your email and I’ll provide access.

Thank you, but I respectfully decline. First, I see that I can already access the app via the URL you provided in your initial post here. Second, there is no way I would run an unvarified app with this level of control on my own system, and I suggest that no one else do so, either. The reason I asked about the public availability of the source code is precisely to provide me and others a higher level of confidence in your work before letting it do what it will with potentially our most personal data.

After reviewing it, I’d appreciate it if you could share your findings publicly — whether you noticed any potentially unsafe parts in the code or anything that could indicate malicious intent, or confirm that nothing concerning was found.

This would do me, you, and, most importantly, your target audience no good at all, even if I were to agree to your proposal. As I am a single anonymous and essentially completely random person on the Internet, my mere word does (and should) inspire no more confidence in your prospective users than does your own. If you are as familiar with the project as no doubt a project's author would be, and if you are as confident in the security and privacy that your hand-written application aims to provide to its users, I see no real reason to insist that it remain private. Who knows? You might even get some contributors to help make the experience even better.

Finally, at the risk of roaming into slightly more personal territory, I want to address briefly an admirable point you raised above:

I’d also like to add that the purpose of this project is to help people who have difficulty interacting with computers and digital interfaces.

That is exactly why I, and maybe others, have concerns. These days, it is all too simple to have a chat bot code up a minimally functional app over a weekend, advertise that app to those who may already both be highly dependent upon the correct functionality of, and face significant challenges in utilizing, their technology, and ultimately cause those people potentially enormous harm. To be clear, I am not saying that this is what you're trying to do. What I am saying is that it's on the developers of these sorts of invasive programs to give us solid, substantive evidence that their products are safe before expecting us to install and run them, and I would be glad indeed to see such evidence provided here.