Dear Reader,

welcome to the August 2025 newsletter. We have a big agenda that could go through most of the year.

 Earned Media

 First, some housekeeping.  You will notice a new page on the site called “Industry News”.  This is where we will be publishing news releases going forward but there are some requirements for companies willing to see their coverage there. First, it is a quid pro quo, but without having to pay a quid for it. We keep being told that paying for play on our pages is a no-go because companies are dedicated to “earned media.” 

 The problem is few people do anything to actually earn that media.  So we have come up with a compromise: you have to promote our magazine on your site or your client’s site.  We will provide a widget saying “Support independent journalism.  Read Cyber Protection Magazine” on your front page with a link to our front page.  You can also link to the page with your release, of course, but the requirement is a front-to-front placement.

 Bypassing search

 There is a reason for this move.  Because search engines are adding AI assistants at the top of pages, links to other pages are disappearing and web traffic is crashing for everyone.  Google is planning on rolling out new advertising programs to pay for link placement in the assistant, but that’s not earned media then, is it?  Our program bypasses Google links altogether and while there isn’t the reach of Google, it creates a more symbiotic relationship.  You know humans are looking at your content.

 Curated content partnership

 Speaking of that, the “breaking news” for us is we are entering into an agreement with Spread (www.joinspread.com) to become one of the first targeted tech magazines featured on the site.  Spread is a social-media adjacent platform that allows users to share content from trusted sources.  No algorithms, no bots.  Users have to prove their humanity. 

 And now for the coming news.

 For better or worse we are in the AI age and our Chief Editor Lou Covey acknowledges that fact as reluctant as he is to embrace it enthusiastically, he does recognize potential benefits. However, he does not abandon his concerns and our next special issue will be focused on these questions.

•   Can we build it? There is no question that generative and agentic AI platforms can be built.  They already have.  But we are at the breaking point of capacity.  Massive new data centers are being planned and the power industry claims it can handle the demand for electrical generation, but it will result in higher energy costs for all users and the potential for community brown outs to keep the power-hungry centers running. Renewable energy with massive storage is one solution.  Small nuclear plants is another.  Both require buildouts over the next 5-10 years.

•   Can we afford to build it? (Each 10 MW nuclear plant comes with a $1.5 billion price tag.  Goldman Sachs predicts the need for an additional capacity of 90,000 MW by 2030 but current plans will only reach 10 percent of that plan. We are talking about trillions of dollars of investment just in power.

•   Can if perform as advertised? The cracks in generative/agentic AI capabilities formed last year are getting wider.  ChatGPT 5.0 may be the peak capability of LLMs.  Companies that laid off thousands of workers to replace them with AI agents are finding it necessary to rehire many more.

•   In 2023, IBM laid off 8,000 HR staff, replacing them with its AI system, AskHR achieving a 94% automation rate, but it couldn’t handle the 6% of cases that required empathy, ethical judgment, or emotional nuance. IBM had to reinvested its savings from the layoffs to hire creative, strategic, and human-centric roles such as engineering, client engagement, and marketing. So no financial benefit.  They aren’t the only company experiencing buyer's remorse and they won’t be the last.

•   Can we protect it? This is where the security aspect pops in.  Infrastructure security in the world is the worst of any sector.  With all the money needed to build and power this new age, there won’’t be much in the budget for security. Imagine OpenAI getting shut down with a ransomware demand of $1 billion.

 •   Is there an alternative? The cybersecurity industry seems to be leading the way to adoption of two new engineering disciplines: small language models and outcome engineering.  In the former, creating AI on limited and vetted content seems to create better outcomes for security (and privacy).  In the latter, security engineers are being deployed to answer the question, “What if it all goes wrong?”  We’d like to hear how that’s working out.

 So, all call out to all of our readers: put your thinking caps on and come up with answers.  We even want opposing viewpoints.  No more than 750 words, no plagiarized/mediocre work (a dead giveaway for AI generation), and no promotional copy.  We're also looking to interview several people for podcasts.  All interviews will be recorded for potential publication.  Contact Lou through the magazine website or to his email at [email protected].

 Also, don’t ask if we want to receive a contributed article.  That’s one of the things we are asking for.  Just send it and make sure it fits the guidelines.

Enjoy Reading

Lou, Joe and Patrick