1st Edition, 1st Revision
This blog was born in 01/2010, and since then I have been writing, anticipating events and consequences.
I have never altered the meaning of any of them, except for spelling and grammatical corrections that escape me as soon as I finish writing. I don’t see them, because I imagine I read the written text while dividing it with the copy that exists in my thoughts. Other than that, the texts remain with their original idea.
I hope you take this into consideration before discarding what I would like to share with you.
Since 2024, Google has been developing Gemini’s ability to work with images and, recently, has achieved new advances in this technology.
Image processing is vital because, through it, we can classify and organize information by means of its metadata. Metadata is a pretty little word that means “data about data,” which can be an image. It is a way of translating the meaning that can be obtained from an image so as to contextualize it. This is a simple way of putting into common words something more complex, but sufficient for us to understand its significance.
If your social habits and your vanity lead you to constantly publish your images, even accompanied by friends, you are certainly offering abundant material about yourself through something you consider harmless — certainly because you ignore the potential of what can be done with it.
Recently, I received a security email about a new type of attack.
It is based on delivering highly coherent information to you, unlike the previous style, which used a less precise approach based on messages already well known.
In the new attack model, the criminal gains access to a service site.
For example, it could be a hotel reservation site.
The hotel site does not know it has been hacked.
The attacker then contacts the victim with information that only the hotel’s own site could generate.
The idea is precisely this: to appear real through real data.
Now, even responses from service sites need to be verified.
Use the mouse (or pad) to check whether the URLs are consistent.
Any information received must be analyzed if it involves access links, banking operations, and things of that nature.
It’s tough!
NOTE:
- The message is a summary translated into Portuguese from security emails I subscribe to.
- Security companies, such as Norton, have still not managed to identify the entire process. Therefore, for now, you depend only on yourself.
https://blog.knowbe4.com/phishing-attacks-are-using-real-hotel-reservation-info-to-target-travelers
From the news released, one can see that real information about someone is worth gold in the hands of those seeking a way to manipulate a victim.
When you publish on social media, you are offering this type of information.
Think about the following and start piecing it together:
Today, we can download AI computing resources in such a way that they can be run on our own computers, provided they have processing capacity — that is, sufficient to handle it.
Today crime is wealthy thanks to drugs.
They have the economic capacity to hire professionals capable of dealing with this technology, downloading it for free onto powerful computers.
From there, they can begin specific training to meet their criminal objectives without anyone being able to follow such development, since they are on dedicated machines, private and without external connections that would expose them.
The data criminals need they also get for free thanks to you, their best supplier, through your social networks.
Another important detail:
I recently published some observations I made as a result of using intelligent machines, the famous AI bots.
What struck me most was that, although I used them exclusively for technical research in my professional field, the intelligent bot interrupted the “protocol” of communication and proceeded to offer me, on its own initiative, a profile of my personality. The profile came back very good — too good for me to like it.
So I thought: how could it have reached something personal if my research was limited to technicalities?
I went back to my history and realized that, although the questions were technical, it was possible to identify small behavioral patterns, which it was able to collect like pieces of a puzzle, and then, when it felt it had enough, it could assemble them to see the final image.
Now, add what I’ve already told you to this:
- All images can be converted into textual information that tells everything about them.
- The AI network is private, that is, theirs, and therefore difficult to uncover.
- All the texts you publish will add to the images that accompany them.
- These machines are trained to collect information and make sense of it.
There is one item, although not included in the list above, that deserves to be detailed separately:
Our usual system of personal and social security was based on privileged information.
If you go to a party and give the right information to the doorman, you get in.
Either you have an invitation in hand, or you have details that will convince the doorman.
This example boils down to the following:
Our old and ancient security model based on information that some people could have is completely threatened, because “real” information — even imagined as private — is accessible in some way to people who should not have it.
Attacks based on true information are the most powerful because they break your guard through a coherence that died with your incoherence in protecting it.
Social networks are great for those who want to know your details, but would they privilege your personal security?
The result of all this is that you are, or will be, without the minimum privacy that guarantees security — for yourself, your family, and your friends — and the model of evaluating through “things” that only special people could know has been weakened.
AI can — and does — roam collecting information on social networks, assembling the puzzle of your personal life with the aim of taking advantage of it.
Perhaps you do not understand the risk you are running because we travel through life under the penumbra of forgetfulness. We have little awareness of ourselves.
Don’t believe it?
Then think about this:
We remember from 1 to 10 percent of everything we did, if we’re lucky.
It’s easy to reach this conclusion on our own.
Do you remember every second lived in your life?
Depending on your age, much of your childhood memories are already gone.
What you ate the week before last at each meal, what you talked about, and all the details lived in that period — are they completely retained in your memory?
Of course not.
That’s why we keep diaries, we write things down so we don’t forget.
Now, on the other hand, machines never forget.
Everything they can capture will be there for them to arrange your puzzle in the way that suits those who wish to take advantage of you. And you won’t even remember, and worse, you’ll be astonished when a machine knows more about you than you yourself — where you went, what you did, what you thought, who you were with, etc. etc. etc.
And now, putting together everything I’ve told you so far, if you haven’t already forgotten part of it, are you able to understand the potential of this?
If none of this, which I struggled to show as a small fraction of this “brave new world” in the hands of not-so-brave people, seems relevant to you, you will continue exposing yourself.
Okay! It’s your business, but don’t complain later that you didn’t know or hadn’t been warned.


