Stratton | AI Cybersecurity in Practice | E-Book | www.sack.de
E-Book

E-Book, Englisch, Band 3, 239 Seiten

Reihe: Plain-English Cybersecurity

Stratton AI Cybersecurity in Practice

How Artificial Intelligence Changes Threat Detection, Phishing Defense, and Enterprise Cybersecurity Operations Today
1. Auflage 2026
ISBN: 979-8-90498-474-8
Verlag: PublishDrive
Format: EPUB
Kopierschutz: 0 - No protection

How Artificial Intelligence Changes Threat Detection, Phishing Defense, and Enterprise Cybersecurity Operations Today

E-Book, Englisch, Band 3, 239 Seiten

Reihe: Plain-English Cybersecurity

ISBN: 979-8-90498-474-8
Verlag: PublishDrive
Format: EPUB
Kopierschutz: 0 - No protection



Something shifted. A colleague forwards you an email that sounds exactly like your CFO. Your board asks what you're doing about AI, and you realize half that question is really about security. You already run a security function - and you can feel the job changing under your feet.

Book 3 of the Next-Gen Cybersecurity Playbooks series is the working guide to AI in cybersecurity today. Follow the fictional Sentinel Ridge AI team through ten chapters that treat AI as both a threat and a tool - with playbooks you can actually run.

Inside you will learn:

• Why AI changed cybersecurity forever - from two weeks to one afternoon, what AI actually added
• Speaking AI fluently - the well-read intern, two families of AI, prompts, RAG, guardrails, evaluations
• AI-generated phishing and voice fraud - why old red flags stopped working, the new pipeline, process controls that survive good lures
• Deepfakes in the workplace - high-risk meeting archetypes, in-meeting identity checks, the ten-second protocol
• Adversarial ML and model poisoning - data poisoning, prompt injection, RAG leakage, defense layers
• AI-assisted detection and triage - the pipeline, what AI can safely close, what humans still own
• AI in threat hunting and response - AI-guided hunting, incident response with AI assist, what to never auto-execute
• Securing the AI your employees use - shadow AI, data that leaves the building, the AI AUP, shadow-to-sanctioned migration
• Evaluating AI security vendors - how to tell substance from theater
• The credible next decade - autonomous SOCs, AI-native identity, the arms race

Ten chapters. Practitioner-first. The AI cybersecurity playbook for the person who has to ship it.

Stratton AI Cybersecurity in Practice jetzt bestellen!

Autoren/Hrsg.


Weitere Infos & Material


Why AI Changed Cybersecurity Forever


The Wire That Started at 9 A.M.


On a Tuesday morning in late spring, Dr. Amara Okafor sat across a conference table from the operations director of a ninety-person accounting firm, and neither of them wanted to be the first to break the silence. The director had a printed bank confirmation in front of him. It showed a wire for three hundred forty thousand dollars, sent to a vendor account that belonged to no one his firm had ever done business with. His controller, a careful woman with twelve years at the firm and a reputation for double-checking everything, had authorized it herself at 9:14 in the morning after what she believed was a completely normal exchange of emails, a short phone call with a voice she recognized, and a follow-up text that arrived exactly when she expected it.

Amara opened her laptop and turned it so he could see the reconstructed attacker workflow her team had put together over the previous forty-eight hours. She walked him through the three AI tools the attackers had almost certainly used. The first was a general-purpose language model that had generated a first draft of the vendor email, complete with a plausible invoice number and a reference to a real project the firm had posted about on its own website three weeks earlier. The second was a voice cloning service that had taken roughly thirty seconds of the firm's managing partner from a public podcast interview and turned it into a passable clone that could speak any script in real time. The third was a small piece of automation glue that stitched the email, the follow-up text message, and the timed voice call into a single choreographed sequence that arrived within an hour of the initial contact.

You could see the director's face change as the timeline compressed before him. He had walked into the meeting assuming the attackers had spent weeks watching his firm, learning the controller's habits, studying invoice formats, and rehearsing the impersonation. That was the story that made sense inside the mental model he had built from a decade of security awareness training. What Amara was showing him instead was a workflow that a moderately skilled operator could have executed in a single afternoon, using tools that cost less than a good dinner and required no meaningful technical expertise beyond the ability to follow a tutorial. By the time she closed the lid, he had already started drafting a text to his board.

This book is about that moment, and the thousands of moments like it happening every week inside firms of every size. It is not a book about the end of security or the triumph of the machines. It is a book about a real, measurable shift in how attacks are produced and how defenses must respond, written for the people whose job is to make good decisions in the next twenty-four months. You will meet Amara again, along with four of her colleagues at Sentinel Ridge Security, throughout the chapters that follow. Sentinel Ridge is fictional, a composite firm assembled from patterns the author has seen across dozens of real engagements, and every client story in this book is a composite as well. The point is never the specific firm. The point is the mechanism.

A quick word about the controller in that opening scene, because the temptation to blame her is exactly the wrong instinct to bring into this book. She didn't fall for the attack because she was careless, and she didn't miss any red flag the training program had taught her to catch. She followed the procedure she had been given. She confirmed the request through a second channel; she recognized a voice she had heard on a dozen prior calls, and she matched the request against a project she knew was in motion. The attackers had done their homework, and their homework was cheap. Any narrative that ends with the reader treating her, or the people in the equivalent seat inside their own firm, as the weakest point in the system is a narrative that will lead to bad decisions. The reader is a partner in defending the firm, not the problem you are trying to work around, and the controls this book describes are built on that premise.

One more note before the mechanism. If you have picked up this book expecting a tour of the state of the art in AI research, you will be disappointed on purpose. The author has read that literature, and it is genuinely interesting, but it is not what your firm needs from you in the next eighteen months. What your firm needs is a working mental model of what has actually shifted, a clear sense of which shifts you can act on and which are still speculative, and a plan that sequences changes to your program in an order that respects your budget cycle and your team's capacity. That is what the pages ahead try to give you.

Before you go further, it is worth naming the thing this book will not do. It will not tell you that AI has invented brand new categories of cybercrime out of thin air. It has not. Phishing existed before large language models, wire fraud existed before voice cloning, and insider mistakes existed before anyone had heard the phrase prompt injection. What has changed is the tempo, the cost, and the polish. When a mediocre attacker can produce something that used to take a skilled operator two weeks in an afternoon, the threat model changes even if the underlying moves look familiar. This chapter makes that argument, and the rest of the book unpacks it.

From Two Weeks to One Afternoon


The single most important change in the last three years is not a new technique. It is a new tempo. To feel it, put two versions of the same attack side by side and watch the clock.

Picture a 2019 attempt against a firm that looks a lot like the one Amara visited. An attacker decides to target the controller. He spends the first three or four days reading LinkedIn profiles, pulling old press releases from the firm's website, and cross-referencing names with public court filings to find a vendor relationship worth impersonating. He spends another two days drafting emails, running them through free translation tools, and rewriting the awkward phrases by hand because his English is uneven and he knows the controller will notice. He spends a day setting up a look-alike domain and configuring the mail server so the messages don't immediately land in the spam folder. He spends another day trying, and mostly failing, to reach the controller by phone using a scratchy internet call, hoping his accent and the poor audio will be blamed on a bad connection. By the end of the second week, he has either succeeded, given up, or been caught. The whole operation has cost him time he could have spent on ten other targets, and the finished product is a middling piece of social engineering that a well-trained employee can catch.

Now picture the 2026 version. The same attacker opens a general-purpose language model in a browser tab, pastes the firm's about page into the prompt, and asks for a plausible vendor email that references a recent project. He gets three drafts in under a minute and picks the one that reads most like a real invoice cover note. He opens a second tab, uploads a thirty-second clip of the managing partner from a public podcast, and generates a voice clone he can drive from a keyboard. He opens a third tab that offers a small automation service, feeds it the email, a follow-up text message, and a script for the phone call, and tells the service to release them in a specific order over the next hour. He tests the whole choreography once against a burner account of his own, tweaks a phrase, and then aims it at the controller. The clock, from cold start to live attack, has moved from about eighty hours of skilled human labor to about three hours of casual clicking. He can now run twenty of these campaigns in parallel, at roughly the same cost, and if nineteen of them fail, he still comes out ahead.

Diagram 1.1.

That compression is the story. Everything else in this chapter follows from it. When the marginal cost of running a well-produced social engineering campaign drops by an order of magnitude, three things happen at once. The volume of attempts goes up, because targets that were not economical to hit before are economical now. The quality of the average attempt goes up, because the floor of what a low-skill attacker can produce has risen to something a careful employee can no longer reliably distinguish from a real message. And the shape of the attempt gets more coordinated, because it is now trivial to stitch together email, text, and voice into a single narrative that arrives in a believable order.

You can feel the second-order effects in the training programs that used to work. For a decade, the standard advice for spotting a phishing email was to look for bad grammar, generic salutations, and mismatched sender addresses. That advice was correct when it was written, and it is not correct now. AI-generated lures pass the grammar test the first time they are drafted. They use the recipient's real first name and the real name of the project they are pretending to reference. They arrive from domains a casual glance will not catch, and within the hour they're followed by a text and a voice call that reinforce the story. When you keep telling employees to look for red flags that no longer show up, you train them to trust messages they should treat with more suspicion, not less.

The compression also changes what a mistake looks like on the defender's side. In the 2019 world, if you missed the signal for two days, you probably still had time to notice the outbound wire request before it cleared. In the 2026 world, the whole attack lands inside a single business morning, and the wire is out of the account before the compliance team has finished their first cup of coffee. That is not a call for panic. It...



Ihre Fragen, Wünsche oder Anmerkungen
Vorname*
Nachname*
Ihre E-Mail-Adresse*
Kundennr.
Ihre Nachricht*
Lediglich mit * gekennzeichnete Felder sind Pflichtfelder.
Wenn Sie die im Kontaktformular eingegebenen Daten durch Klick auf den nachfolgenden Button übersenden, erklären Sie sich damit einverstanden, dass wir Ihr Angaben für die Beantwortung Ihrer Anfrage verwenden. Selbstverständlich werden Ihre Daten vertraulich behandelt und nicht an Dritte weitergegeben. Sie können der Verwendung Ihrer Daten jederzeit widersprechen. Das Datenhandling bei Sack Fachmedien erklären wir Ihnen in unserer Datenschutzerklärung.