Tag Archives: security

Why Replacing Your Employees with VLM NPE Bots Won’t Defeat Social Engineering

(Scammed bot finger picture from Imagen 3)

Your cybersecurity firm can provide the most amazing protection software to your clients, and the clients still won’t be safe.

Why not? Because of the human element. All it takes is one half-asleep employee to answer that “We received your $3,495 payment” email. Then all your protections go for naught.

The solution is simple: eliminate the humans.

Eliminating the human element

Companies are replacing humans with bots for other rea$on$. But an added benefit is that when you bring in the non-person entities (NPEs) who are never tired and never emotional, social engineering is no longer effective. Right?

Well, you can social engineer the bot NPEs also.

Birthday MINJA

Last month I wrote a post entitled “An ‘Injection’ Attack That Doesn’t Bypass Standard Channels?” It discussed a technique known as a memory injection attack (MINJA). In the post I was able to sort of (danged quotes!) get an LLM to say that Donald Trump was born on February 22, 1732.

(Image from a Google Gemini prompt and response)

Fooling vision-language models

But there are more serious instances in which bots can be fooled, according to Ben Dickson.

“Visual agents that understand graphical user interfaces and perform actions are becoming frontiers of competition in the AI arms race….

“These agents use vision-language models (VLMs) to interpret graphical user interfaces (GUI) like web pages or screenshots. Given a user request, the agent parses the visual information, locates the relevant elements on the page, and takes actions like clicking buttons or filling forms.”

Clicking buttons seems safe…until you realize that some buttons are so obviously scambait that most humans are smart enough NOT to click on them.

What about the NPE bots?

“They carefully designed and positioned adversarial pop-ups on web pages and tested their effects on several frontier VLMs, including different variants of GPT-4, Gemini, and Claude.

“The results of the experiments show that all tested models were highly susceptible to the adversarial pop-ups, with attack success rates (ASR) exceeding 80% on some tests.”

Educating your users

Your cybersecurity firm needs to educate. You need to warn humans about social engineering. And you need to warn AI masters that bots can also be social engineered.

But what if you can’t? What if your resources are already stretched thin?

If you need help with your cybersecurity product marketing, Bredemarket has an opening for a cybersecurity  client. I can offer

  • compelling content creation
  • winning proposal development
  • actionable analysis

If Bredemarket can help your stretched staff, book a free meeting with me: https://bredemarket.com/cpa/

Now I’m Just Playing with Google Gemini

I asked Imagen 3 to help me illustrate nth party risk management.

Where you are connected with everyone to whom your connections are connected.

But I wanted to illustrate third-party risk management in a clean way. Back when AIDS became a sad feature of our lives in the 1980s, the description of how it spread from person to person could get a little graphic.

Why Does TPRM Fail? Not Because of the TPRM Software Providers.

For years I have maintained that the difficulties in technology are not because of the technology itself.

Technology can do wonderful things.

The difficulties lie with the need for people to agree to use the technology.

And not beg ignorance by saying “I know nothing.”

(Image of actor John Banner as Sgt. Schultz on Hogan’s Heroes is public domain.)

Case in point

I just saw an article with the title “TPRM weaknesses emerge as relationship owners fail to report red flags.”

Unlike some clickbait-like article titles, this one from Communications Today succinctly encapsulates the problem up front.

It’s not that the TPRM software is failing to find the red flags. Oh, it finds them!

But the folks at Gartner discovered something:

“A Gartner survey of approximately 900 third-party relationship owners…revealed that while 95% saw a third-party red flag in the past 12 months, only around half of them escalate it to compliance teams.”

Among other things, the relationship owners worry about “the perceived return on investment (ROI) of sharing information.”

And that’s not a software issue. It’s a process issue.

wildebeests on a stairway, young to old, with the oldest wildebeest possessing a trophy
Wildebeest maturity model via Imagen 3.

No amount of coding or AI can fix that.

And this is not unique to the cybersecurity world. Let’s look at facial recognition.

Another case in point

I’ve said this over and over, but for U.S. criminal purposes, facial recognition results should ONLY be used as investigative leads.

It doesn’t matter whether they’re automated results, or if they have been reviewed by a trained forensic face examiner. 

Facial recognition results should only be used as investigative leads.

Sorry for the repetition, but some people aren’t listening.

But it’s not the facial recognition vendors. Bredemarket has worked with numerous facial recognition vendors over the years, and of those who work with law enforcement, ALL of them have emphatically insisted that their software results should only be used as investigative leads.

All of them. Including…that one.

But the vendors have no way to control the actions of customers who feed poor-quality data into their systems, get a result…and immediately run out and get an arrest warrant without collecting corroborating evidence.

And that’s not a software issue. It’s a process issue.

No amount of coding or AI can fix that.

I hope the TPRM folks don’t mind my detour into biometrics, but there’s a good reason for it.

Product marketing for TPRM and facial recognition

Some product marketers, including myself, believe that it’s not enough to educate prospects and customers about your product. You also need to educate them about proper use of the product, including legal and ethical concerns.

If you don’t, your customers will do dumb things in Europe, Illinois, or elsewhere—and blame you when they are caught.

Illinois, land of BIPA. I mean Lincoln.

Be a leader in your industry by doing or saying the right thing.

And now here’s a word from our sponsor.

Not the “CPA” guy again…

Bredemarket has openings

There’s a reason why this post specifically focused on cybersecurity and facial recognition.

If you need product marketing assistance with your product, Bredemarket has two openings. One for a cybersecurity client, and one for a facial recognition client. 

I can offer

  • compelling content creation
  • winning proposal development
  • actionable analysis

If Bredemarket can help your stretched staff, book a free meeting with me: https://bredemarket.com/cpa/

Bredemarket has openings. Imagen 3 again.

The Military, Cyberattacks, and Maturity

Everyone knows that cyberattacks don’t just target private organizations. They also target governments, particularly aiming for agencies that either deal with a lot of money (unemployment agencies) or contribute to defending a country (military, homeland security).

The Chief Information Officer of the U.S. Department of Defense has a vested interest in preventing cyberattacks, not only against DoD, but against its third-party suppliers, which are the subject of today’s acronym, DIB (defense industrial base).

And if you’ve followed along in the Bredemarket blog lately, you know that a key component of preventing cyberattacks is raising your organization’s process maturity in the cybersecurity realm.

And yes, there’s a maturity model and a certification for that, the Cybersecurity Maturity Model Certification, or CMMC.

Cybersecurity is a top priority for the Department of Defense (DoD). The defense industrial base (DIB) faces increasingly frequent, and complex cyberattacks. To strengthen DIB cybersecurity and better safeguard DoD information, the DoD developed the Cybersecurity Maturity Model Certification (CMMC) Program to assess existing DoD cybersecurity requirements.

It’s no surprise that the CMMC incorporates multiple levels, in this case three of them.

  • Level 1: Basic Safeguarding of FCI (Federal Contract Information)
  • Level 2: Broad Protection of CUI (Controlled Unclassified Information)
  • Level 3: Higher-Level Protection of CUI Against Advanced Persistent Threats

And not only is there a maturity model certification for the defense industrial base, but there’s a conference to help everyone out. After all the geeks celebrate May the Fourth Be With You day, some of the geeks will continue to celebrate on May 5, the date of the fifth annual CMMC Day. Party on.

Also see Biometric Update’s article, as well as NIST SP 800-171 Rev. 3, Protecting Controlled Unclassified Information in Nonfederal Systems and Organizations.

And if you need product marketing assistance with your cybersecurity product, Bredemarket has an opening for a cybersecurity client and can help with compelling content creation, winning proposal development, and actionable analysis. Book a call: https://bredemarket.com/cpa/ 

(Military wildebeest image from Imagen 3)

Startups Entering the Wonderful World of Process

I’ve talked about governance and maturity models before in regards to cybersecurity. The complicating factor is that companies with little process maturity are flung into the world of standards and auditors. 

For example, I was not initially part of the process team when the former seat-of-the-pants Printrak had to play CMM catch up with our new corporate overlord Motorola. But it was a bruising experience.

These days you have a lot of startups, not owned by multinationals, that are required by large customers and governments to comply with some standard or another. Winging it is not an option; winging it is failure. Or, in process-speak, winging it can result in a high statistical probability of a large number of adverse  findings.

Vanta wants to help.

Its early April “Guide to working with auditors: Best practices for startups” contains several suggestions. 

  • One is to engage with auditors early so that you become familiar with each other.
  • However, you should NOT give auditors access to your data early. Wait until you are ready. Assuming your data is in a Vanta instance:

“If you’re still finalizing controls in Vanta, granting early access could cause confusion. However, some firms prefer early access for familiarization—as long as they don’t start testing prematurely.”

Vanta’s guide is at https://www.vanta.com/resources/guide-to-working-with-auditors-for-startups

(Wombat image via Imagen 3)

PS to cybersecurity product marketers

Are you getting YOUR product’s message out? Or is a stretched team holding you back from creating stellar marketing materials? 

Bredemarket has an opening for a cybersecurity client and can help with compelling content creation, winning proposal development, and actionable analysis. Book a call: https://bredemarket.com/cpa/ 

Why Invela TPRM?

During my three months working with a third-party risk management (TPRM) client, I never heard anyone mention Invela.

Perhaps with reason. Although LinkedIn says the company was founded in 2024, it didn’t post its first blog until April 20, 2025, or its first LinkedIn posts until April 21.

But the second blog post, dated April 21, is the one that matters.

“Invela has officially launched a transformative network to bolster consumer protection and foster innovation within the open banking ecosystem. The Invela Network, developed in collaboration with industry-leading specialist partners, promises to revolutionize how financial institutions manage third-party risk…”

The post goes on to cite the Consumer Financial Protection Bureau (CFPB), but…well…that’s nice.

Invela’s TPRM solution specifically targets the open banking segment of the financial services industry. Open banking, featuring companies such as Plaid, Kong, and Camunda (among others), facilitates the interchange of financial data, rather than keeping it within each bank’s walled garden.

Which of course increases risk.

Hence companies such as Invela.

I was unable to find a “why” story for Invela that compared to the why story I previously found for Ubiety Technologies. Obviously the Invela people never read my book.

However, the principals at Invela come from companies such as Mastercard (although I could find no information on Invela’s CEO Steve Smith). But the Invela leadership team presumably knows their market. We will see if they know their marketing.

Which reminds me…if you need help with your cybersecurity product marketing, Bredemarket has an opening for a cybersecurity client. I can offer

  • compelling content creation
  • winning proposal development
  • actionable analysis

If Bredemarket can help your stretched staff, book a free meeting with me: https://bredemarket.com/cpa/

A Mature Approach to Artificial Intelligence-Powered TPRM Automation

Deloitte conducts regular surveys on third-party risk management (TPRM), and just concluded a survey on (English warning) “the rise of AI in TPRM to maximise opportunities while managing the risks.”

One of the key findings:

“Despite low maturity levels, leadership teams are ambitious about embracing intelligent automation, while managing both the risks of AI in their organisations and those arising from third-party AI usage.”

I’ve talked about maturity levels before and their importance in cybersecurity. While ad hoc approaches to TPRM just won’t cut it in terms of protection, a managed or defined level or better will yield a positive return on investment.

(Imagen 3)

And one more thing…

The formal announcement is embargoed until Monday, but Bredemarket has TWO openings to act as your on-demand marketing muscle for facial recognition or cybersecurity:

  • compelling content creation
  • winning proposal development
  • actionable analysis

Book a call: https://bredemarket.com/cpa/ 

Frame, Assess, Respond, and Monitor (FARM) in Third-Party Risk Management

I just listened to a third-party risk management (TPRM) Mitratech webinar about NIST cybersecurity frameworks, hosted by OCEG, which talked about a farm.

No, they’re not planting corn at NIST’s Gaithersburg headquarters.

(At least I don’t think so. I haven’t been there since early 2009, back when Motorola and Safran people couldn’t talk about the possible acquisition. We did anyway. But I digress.)

Back to TPRM. In Mitratech’s case, FARM stands for “frame, assess, respond, and monitor.”

Here’s how Mitratech introduced the topic in a 2022 post:

NIST SP 800-53 is considered the foundation upon which all other cybersecurity controls are built. With SP 800-161 Rev. 1, NIST outlines a complementary framework to frame, assess, respond to, and monitor cybersecurity supply chain risks. Together, SP 800-53 and supplemental SP 800-161 control guidance present a comprehensive framework for assessing and mitigating supplier risks.

If you visit the latest (as of 2024) update to SP 800-161, you can find NIST’s explanation of the FARM in Appendix G. The three referenced levels in the quote below are the enterprise, mission, and operations levels.

The first approach is known as FARM and consists of four steps: Frame, Assess, Respond, and Monitor. FARM is primarily used at Level 1 and Level 2 to establish the enterprise’s risk context and inherent exposure to risk. Then, the risk context from Level 1 and Level 2 iteratively informs the activities performed as part of the second approach described in The Risk Management Framework (RMF). The RMF predominantly operates at Level 3 [SP80037], – the operational level – and consists of seven process steps: Prepare, Categorize, Select, Implement, Assess, Authorize, and Monitor.

Briefly:

  • Frame establishes the context.
  • Assess is the risk assessment itself.
  • Respond is where the assessors communicate the results of the assessment and propose mitigations and controls.
  • Monitor is compliance verification and continuous monitoring.

Section G.2 of the document includes much, much more detailed definitions of the FARM elements, should you be interested. I’d provide those details myself, but then I fear I’d have to say to you, “Sorry if I’ve stayed too long.”

If Your Identity System Only Manages People, It Is Flawed

This is painful, but it has to be done.

I’ve spent 30 years working with the identities of PEOPLE and ensuring that all PEOPLE accessing a system are properly identified.

In other words, leaving a huge GAPING security hole.

Look at what Okta is doing;

“[N]ew Okta Platform capabilities…help businesses secure AI agents and other non-human identities with the same level of visibility, control, governance, and automation as human ones. The Okta Platform will now bring a unified, end-to-end identity security fabric to organizations for managing and securing all types of identities across their ecosystem, from AI agents to API keys to employees.”

I think that “unified” will take the place of “trust” as the identity buzzword. Thankfully.

If you’re only selling biometrics, or maybe biometrics and ID cards, where will your customers go to get the rest of their systems? Or will you just be a commodity supplier to the companies that provide the REAL systems?

(Unified security AI picture from Imagen 3)