Ahad, 30 Ogos 2026


Think about the last thing you asked AI.

Maybe it was about work.

Maybe relationship advice.

Maybe you pasted an email and asked, "What does this person actually mean?"

Maybe you uploaded a document and said, "Summarise this for me."

Or maybe at 2AM you asked something so random and personal that you would never type it into Google. 😅

That's the interesting thing about AI.

We don't use it like search. We talk to it.

Google Gets Keywords. AI Gets Context.

With Google, we learned to keep things short.

"Best laptop 2026."

"Cheap flight Tokyo."

"How to resign professionally."

AI feels different.

We explain the whole story.

Who said what. What happened yesterday. What we're worried about. What we're planning to do next.

Sometimes we paste emails, reports, screenshots, code, meeting notes or personal conversations because more context usually gives us a better answer.

And that's exactly why we should think differently about privacy.

The Better We Want AI To Be, The More We Tell It

AI becomes useful when it understands context.

Want it to rewrite an email? Give it the email.

Want career advice? Tell it about your career.

Want it to analyse a document? Upload the document.

Want personalised recommendations? Tell it what you like.

It feels natural because we're having a conversation.

But somewhere along the way, "using a tool" starts feeling like "talking to someone."

And when that happens, we may share more than we normally would.

The Problem Isn't AI. It's What We Give It.

This doesn't mean we should stop using AI.

I use AI because it's useful.

But before pasting something into any AI service, maybe we need one simple habit:

Think before you paste.

Does this contain confidential company information?

Customer data?

Passwords or credentials?

Someone else's personal information?

Something you wouldn't normally upload to an external service?

AI can only work with the context we give it.

And sometimes we're giving it a lot.

The same principle applies in cybersecurity. In The Real Risk of AI Coding Agents Is Not Intelligence. It Is Permission., the bigger question isn't simply how smart AI becomes, but what information and access we allow it to have.

Maybe AI Needs a New Privacy Habit

We spent years learning not to click suspicious links.

Not to reuse passwords.

Not to share OTPs.

Maybe the AI generation needs another basic internet rule:

Don't tell AI something just because it feels like a private conversation.

Because AI is becoming easier to talk to.

More personal.

More helpful.

More human sounding.

And perhaps that's exactly why we need to remember one thing.

It may feel like a conversation. But it's still technology.

Next
This is the most recent post.
Previous
Catatan Lama