Wednesday, December 25, 2024

How Non Human Identities Outnumber The Human Population By 50:1

 


Back in the day, when I was in graduate school during the Internet Boom, the two big buzzwords back then were “B2B” and “B2C”.  These both stand for “Business to Business”, and “Business to Consumer”, respectively.  Well fast forward from those times to now, where the world of Cybersecurity is now filled with all kinds of techno jargon. 

And with the explosion of Generative AI, the dictionary here expands with a newer one, which is called “M2M”, or “Machine to Machine”.

Now this term has just recently given birth to an even newer one, which is now “NHI”, or “Non-Human Identities”.    These are the Chatbots, Virtual Personal Assistants, and even the Digital Personalities that you engage in instead of speaking with a real human being.

 These kinds of NHIs are now literally dominating the world.  In fact, it has even been cited that they outnumber actual human identities by a factor of 50:1. 

Because when you communicate with an NHI, you are giving away your own personal information and data.  In turn, the Generative AI models that power them are storing them to not only recall this information/data if you were to engage with them, but to also train the algorithms so that they can be fully always optimized.

Apart from all of this being transmitted back and forth, there is now a cry in the Generative AI world that there needs to be set of bet practices and standards that businesses must adhere to if they make use of NHI that is customer facing.  Here are the factors that are driving this movement:

1)     Complexity:

Gone are the days when businesses just had an On Premises Infrastructure.  As I have written about before in previous blogs, there is now a strong movement to go to the Cloud, using a major platform such as that of the AWS or Microsoft Azure.  But, there are still some CISOs who relish the old fashioned ways of doing things, so they opt for a hybrid based approach, which is a combination of both On Premises and the Cloud.  This kind of blend does not make things any easier, in fact, it makes it much more complex for the IT Security team to manage.  If an NHI is created and deployed by using both words, then everybody needs to follow and stick to the same set of rules, especially when it comes to data/information storage and processing.

2)     Automation:

One of the biggest benefits of that Generative AI brings to the table is that it can be used to automate processes and functions.  For example, this can be seen with robotic arms in a car assembly plant, and even in Cybersecurity, it is being used in Penetration Testing and Threat Hunting in the more mundane and routine tasks.  But, there is a key problem here:  You simply cannot rely upon automation 100% (at least in my view).  For example, what if there is no human intervention if a Data Exfiltration Attack were to occ. to an NHI?  Well, the bottom line is that it would go completely unnoticed, until it is way too late.  This debate about whether to completely automate or not is currently one of the biggest debates that is happening and will continue for a long time in the world of Generative AI.

3)     Easy Prey:

An NHI in any form is actually a very easy prey for the Cyberattacker to go after.  They may not attack it directly, but they can very easily go after the connections which they are linked to.  These are often not very secure, and once a Cyberattacker is able to penetrate through just one of these network lines of communications, they can wreak all sorts of havoc in just a short matter of time.

4)     Mergers And Acquisitions:

There are a lot of buyouts that are taking place with Cyber Vendors.  Some of these include the following:

Ø  Authomize being purchased by Delinea.

Ø  Venafi being purchased by Cyber Ark

It should be noted that the above two buyers are Privileged Access Management (PAM) vendors.  The point here is that with all these mergers happening, there is a lot of information and data that is being transferred, especially with the Generative AI models and the NHIs that they power.  It is very necessary here to have a standard checklist that both the buyer and the buyee must abide by to make sure that nothing is leaked out, intentionally or not.

My Thoughts on This:

As for me, I am still very old fashioned.  Although I am a technical writer in the world of technology, I absolutely hate technology.  I like the way we do things.  I am not sure if I am up for all this Generative AI stuff.  For example, I would much rather see my doctor in person, rather than chat with a Digital Personality.

But if I must bend on this, we cannot depend upon Generative AI on its own.  We need to have human intervention here.  As for the best practices and standards, it is about time that Corporate America did something about it.  The Federal Government has done something about it, but it is way too slow to keep up with the rapid changes that are happening in Generative AI.

To use the old proverb, “its going to take a village” for all this to happen.  It will require a hands-on deck cooperation with the private and public sectors, as well as with academia.  But even if we were to produce such a set of best practices and standards, who is going to enforce it? 

The FBI?  The DHS?  These are tough questions that still must be answered, and as a society, we must figure all of this out soon.

No comments:

Post a Comment

Here Comes 2025: The Major Cyber Threats To Happen

  Ok, here we   go, as we fast approach now into 2025, here are the predictions as what the major Threat Variants and Attack Vectors will be...