Claude Is Connecting Directly To Your Personal Apps
- Reference: 0181965366
- News link: https://tech.slashdot.org/story/26/04/24/066231/claude-is-connecting-directly-to-your-personal-apps
- Source link:
> Some of these apps, such as Spotify, already have similar connectors in OpenAI's ChatGPT. Once an app is connected, Claude will suggest relevant connected apps directly in your conversations, like using AllTrails for hike recommendations. Anthropic notes in its blog post announcing the new connectors that, "Your data from [connected apps] isn't used to train our models, and the app doesn't see your other conversations with Claude. You can also disconnect it at any time."
>
> Additionally, Anthropic says "there are no paid placements or sponsored answers in conversations with Claude." When multiple apps seem relevant, Claude will show results from both "ranked by what's most useful." Claude will also ask users to verify before taking actions like making a purchase or reservation using a connected app.
[1] https://www.theverge.com/ai-artificial-intelligence/917871/anthropic-claude-personal-app-connectors
Incorrect! (Score:2)
It only has access if you use it! Or if you allow unfettered access by using a Microsoft product!
Re: (Score:2)
You think other OSs are safe? No.
[1]https://www.theregister.com/20... [theregister.com]
[1] https://www.theregister.com/2026/04/20/anthropic_claude_desktop_spyware_allegation/
Re: (Score:3)
> You think other OSs are safe? No. [1]https://www.theregister.com/20... [theregister.com]
Arch Linux only gets stuff I choose to install installed on it.
[1] https://www.theregister.com/2026/04/20/anthropic_claude_desktop_spyware_allegation/
Re: (Score:2)
> adding personal-service connectors
It's like the miracle of television, reduced to whoring detergent to housewives.
"your data" (Score:2)
""Your data from [connected apps] isn't used to train our models, and the app doesn't see your other conversations with Claude."
Because it's not "your data" it's our data. And they aren't "your conversations" they're our conversations. If you have a problem with that, Claude may"disconnect you at any time."
That isn't the story here. It's much worse (Score:5, Interesting)
Claude is pre-installing *AND CHANGING SECURITY PERMISSIONS OF OTHER APPS* to pre-authorise its plugin/extension on MacOS. In some cases you'll find a fresh install of some Electron or Chromium based browser and you'll magically see Claude's extension already there. It apparently does so without consent or any indication that it does so.
[1]https://www.theregister.com/20... [theregister.com]
[1] https://www.theregister.com/2026/04/20/anthropic_claude_desktop_spyware_allegation/
Re: (Score:3)
What are you complaining about? This is working as designed. How else would Claude be able to free you from all that burdensome thinking and doing?
Re: (Score:1)
If people were "thinking and doing" in the first place we wouldn't have Trump.
Re: (Score:2)
> If people were "thinking and doing" in the first place we wouldn't have Trump.
Nor would we have this ridiculous 2 party system.
Totally legit and believable (Score:2)
Not one exaggerated claim so far from them, best security, guaranteed by the hackerest AI out there, etc. etc.
What a wonderful marketing opportunity (Score:5, Insightful)
This sounds wonderful. Now you can connect this flawless AI agent to your shopping apps and it can shop for you! Just need to confirm the purchase. Now I know 100% that our honorable, ethical and morally righteous retailers and AI creators would never purposely add "thoughtful" items to my cart and ask if I want this ordered. /s
If you thought advertising was bad, just wait until this starts building carts of "useful" items for you and ask for your confirmation. Slashdot crew will be less likely to confirm it but there are a lot of folks that will go, gee, that seems like a good idea let's spend money I hadn't even considered spending on stuff I most certainly don't need but it looks neat.
All this winning is starting to hurt.
Re:What a wonderful marketing opportunity (Score:4, Funny)
"Confirm the purchases"? That is sooooo yesterday. You really need to get with the times. Obviously, you also give Claude access to all your assets, and it will use them and a few nice consumer credits it arranges on top of that to make the purchases for you, with no work on your part required! You got to go all-in on agentic AI to really get the benefits, obviously.
No (Score:3)
I do not let malware onto my devices. But I will be watching the "early adopters" (a.k.a. victims) with a high level of amusement.
hear hear. (Score:2)
Clickbait headline, seems there are lot of these type of confrontational titles on here in recent years. Good journalism does not assume or directly accuse the reader of being a victim.
Better than... (Score:2)
Better than Clouseau connection to your apps...
--JoshK.
Re: (Score:2)
[hand slides off globe]
Full sandbox is the only way forward. (Score:3)
In order to start using Claude (and try other agents) I bought a new development desktop PC.
It's completely isolated from my home network, my data and even production databases. All it has access to is my repos. Deployment is done from a separate machine with no agents on it.
It's the only safe way forward these days, the only way to protect your personal information, photos and other sensitive data from being harvested without your consent.
Re: (Score:2)
And what was it for the last twenty-five years? Seems people say, AI and it's suddenly, privacy, security, etc. Something that should have been done when "database nation" was published in 2000.
Re: (Score:2)
I'm impressed that anyone can afford a new development PC in 2026. I'd need a second mortgage with the price of the RAM, SSD and GPU these days!
Re: (Score:2)
> I'm impressed that anyone can afford a new development PC in 2026. I'd need a second mortgage with the price of the RAM, SSD and GPU these days!
You could get a Mac Mini if you're wanting to experiment with local models or AI agents...
Their M chips with unified shared memory is uniquely set up to work with AI....often for cheaper than the Win alternative.
That being said, the Minis, especially with more than basic RAM are getting hard to get with longer shipping times...due to popularity combined with like
Re: (Score:2)
Yes, Apple's shared RAM model really works for them in the context of running LLMs locally. It's a huge advantage. As you say, not much use for those running other platforms, though.
Autonomy for the Elites (Score:3)
I feel like all of this "let AI make decisions for you" is eroding our autonomy. We will gradually lose our ability to self-direct and plan while being expected to execute what ever AI has committed us to. The elites will be able to have others work AI for them or use it to drive their lifestyle choices so their businesses do not interfere. Or they can completely ignore it. Because the more money you have, the less it matters how much any use of it impacts your decision making. The elites don't need AI to buy them time, that's what they have money for.
Re: (Score:2)
> I feel like all of this "let AI make decisions for you" is eroding our autonomy. We will gradually lose our ability to self-direct and plan while being expected to execute what ever AI has committed us to.
Do you not understand that is the intent?
Why is the future so lame? (Score:1)
We used to dream of a cool future amongst the stars and with hoverboards, now all we actually get are lame agentic LLMs doing something half-random and asking for full permissions over everything everywhere because surely what could go wrong?
Good idea (Score:2)
AI should not be attempting to do what your existing Apps already do.
In fact, it should connect to your calculator app as well. When your request involves math, it should use the calculator, not try to do it itself.
Juxtaposition (Score:2)
Someone funny in a dark way that this story is posted right above the FEC's attempt to control mass surveillance via hardware. This kind of thing makes it absolute clear that one of the core goals of these self-described "AI" systems it to finalize the capture of all PII on everyone and transmit it to centralized storehouses controlled by... who exactly?
"I see now what I did wrong" (Score:3, Funny)
Me: "Claude, I asked you to arrange an exciting birthday party for my wife and invite all her friends. She has just filed for divorce."
Claude: "I see now what I did wrong. Inviting you, her and all your ex-girlfriends to a strip bar was incorrect. Let me fix that for you."
Re: (Score:2)
Or it invites all her current and past secret lovers. Now that would be something. Definitely "exciting"!
Re: (Score:2)
Claude: I see multiple hostile messages and an invite involving your (soon to be ex) wife is scheduled. Found a few suggested hikes via the AllTrails app - should I proceed with creating a day plan?