How much do you trust AI agents?
With the advent of clawdbots, it's as if we've all lost our inhibitions and "put our lives completely in their hands."
I'm all for delegating work, but not giving them too much personal/sensitive stuff to handle.
I certainly wouldn't trust something to the extent of providing:
access to personal finances and operations (maybe just setting aside an amount I'm willing to lose)
sensitive health and biometric information (can be easily misused)
confidential communication with key people (secret is secret)
Are there any tasks you wouldn't give AI agents or data you wouldn't allow them to access? What would that be?
Re. finances – Yesterday I read this news: Sapiom raises $15M to help AI agents buy their own tech tools – so this may be a new era when funds will go rather to Agents than to founders.
Replies
Yeah, I draw pretty hard lines too.
Anything irreversible or deeply personal stays human for me. That includes:
full access to finances (I’ll allow read-only or capped actions at most)
health, biometric, or identity data
private communications where trust or intent really matters
decisions with legal or long-term consequences
AI agents are great for prep, analysis, drafts, and coordination — but not for final authority. I’m fine letting them recommend, not decide, especially when the downside isn’t recoverable.
minimalist phone: creating folders
@alpertayfurr I wouldn't be happy if Clawd would send some rude message to my clients. :DDD
Definitely agree on personal finances and biometric data.
But I also draw a hard line at social media autonomy. I would never give an agent write-access to my LinkedIn or X accounts to post or reply automatically. My online presence is my reputation.
minimalist phone: creating folders
@valeriia_kuna Me as well. I hate it even when someone uses AI content. It feels so fake and synthetic.
@busmark_w_nika Me too! When I see some AI generated posts on social networks, I'm like🙄🙄🙄
I like to polish my texts with AI or translate them, but I don't like AI generic content.
minimalist phone: creating folders
@valeriia_kuna I do the same, but if I do no like it, I will delete it anyway. sooo.
I trust them with tasks, not with judgment.
I’m fine giving repetitive work. Not finances, private conversations, or anything sensitive.
minimalist phone: creating folders
@kashyaprathod That's my approach too :)
minimalist phone: creating folders
@tereza_hurtova We should consider purchasing a separate device where we can run these agents. :D In general, I have trust issues :D
I wouldn't really want to give any personal data to AI or delegate any kind of management to it. For example, i think using AI to structure data or analyze public information is fine. But I'm not sure that I can trust AI to manage anything. And it doesn't matter whether it's managing my calendar, emails, or personal data.
This is quite literally why we built kwAI to enable people selling rather than let AI do the selling.
We let AI find, research, and draft the messages, while the human does the relationship building.
Very good mix.
minimalist phone: creating folders
@ryan_tucker13 can you shar a link pls?
@busmark_w_nika Of course!! https://i-kwai.com
We're launching next month. :)
minimalist phone: creating folders
@ryan_tucker13 thank you, feel free to remind :)
@busmark_w_nika Amazing! Will do! We're launching in less than 2 weeks! :)
We really appreciate the support. You're welcome to join our Herd any time!
I build AI tools and honestly I don't trust my own agents with anything I can't undo. Like I'll let them draft emails all day but actually sending? Nope, always a human in the loop there. The finance thing is wild to me - seen too many hallucination edge cases to let an agent anywhere near real money. I think the trust question really comes down to reversibility. Read-only access to my calendar, notes, whatever - sure. But anything that creates a side effect in the real world needs a confirmation step. The people skipping that are gonna learn the hard way.
minimalist phone: creating folders
@mykola_kondratiuk Will you be launching any of your agents publicly in the future?
Define AI Agent. I put together an offline ai does that count or are you saying trust the ai companies? If I had the choice I would not use ai companies and just use my own but its expensive, complex and the data centres got the convenience. We should build a decentralized ai and then we can buy personal ai node instead of a data center if you ask me.
minimalist phone: creating folders
@jayotis
Having your own or a local solution would be the ideal scenario. But yeah, it would require large amounts of high-quality data, significant computing power, access to APIs and infrastructure, money etc. As you said, it is not cheap :D
@busmark_w_nika :) my isolated ai worked but was slow as molasses and needed careful prompting, so careful I just coded it myself in the end. I didn't have a gpu that had ai drivers so I was stuck with cpu only, looks like ai graphics cards with 32GB VRAM is $1000 and rising(I suspect market manipulation to stop us from doing this) so that is what I meant by expensive.
The decentralized ai would not be expensive. The models are p2p shared and compiled, we share cpu/gpu time(think SETI but ai) when we are not using our personal ai and the hardware itself could be a raspberry pi in your basement so not expensive. When We do use it we can use other people idle cpu/gpu so its like you have better hardware.
The numbers you are seeing with ai as far as infrastructure are a false indication of the true cost because it is all primitive tech. We are seeing the Hit & Miss engine of ai basically. all that process power and ram storage is because the companies are brute forcing a solution, rushing to market, wasting absurd amounts of energy and cpu time to make it look good and usable so someone will subscribe.
long story short, you are right that using ai data centers is not correct it is supposed to be a personal experience.
During my work we stay away from agents and automations too much, we have a very high security policy set up and with a reason, I also think it is still to far away for the basic users, currently its really at a stage where it helps developers, but we should be focussing on making tools that will help people. Offline agents, not hooked up to anything that will help elderly with reminders for medication and excersize, help children with learning disabilities to get up to speed, set up daily plannings, do groceries not everything has to be online and hooked up to heavy AI. Big data and the internet of things have been topics that have been there for decades and we're still not at a level we ought to be..
minimalist phone: creating folders
@harrie_vermeulen In what industry do you work when you cannot use AI or agents?
@busmark_w_nika Just to be sure, we use AI a lot, but only the secured versions, but agents to change sensitive content is no-go, there are too many validation flows in the processes making it too sensitive for agents to take over fully
When it comes to accessing personal data, I'm not keen. For coding things I can't where it's isolated from my system, I'm happy to work with it. I think it will be a long time to build trust with AI. We've seen it at its very early stages of development and it's still massively error prone.
minimalist phone: creating folders
@liam_oscarlena You know... but AI is becoming normalised, indeed, it is here for 3 years publicly. I think that more and more people will give data voluntarily and this will become a norm.