If your AI can write code… it should help secure it, too.

Anthropic just rolled out Claudе Code Security, a new feature designed to scan codebases for flaws and suggest patches.

AI is already great at parsing logs and highlighting anomalies. But stepping into the auditor's shoes to patch code? That requires deep context.

The true test isn't if Claude can find a flaw; it's whether it understands the messy reality of a production environment without hallucinating a "fix" that breaks the build.

Here's how to use Claude Code Security safely:

1️⃣ Extra pair of eyes
→ Run AI scans on every merge and pull request
→ Let it flag risky patterns

2️⃣ Human in control
→ Security engineer or senior Dev reviews each AI fix
→ No auto-merge from AI output

3️⃣ Tie into threat intel
→ Watch CISA Known Exploited Vulns
→ Confirm your codebase isn't using specific vulnerable functions of the CVEs

I extensively use AI for day-to-day work. For example, for threat intel summary, customer email draft, or compliance audit prep. It’s a fantastic junior analyst. But it is always an assistant, not the one signing off on the decisions.

Do you trust AI to patch your production code or not? 🤔

For the #CyberMonday News and CVE alert, see the first comment. 👇

@securediary

Four years of full-scale war. 1,461 days of resilience.

When I served as a SOC Division Chief in the Armed Forces, we prepared for hybrid threats. But the reality of the last four years rewired everything I know about defense.

Living and working in Kyiv, I’ve seen the concept of "Business Continuity" transform from a compliance checkbox into a survival instinct. We don’t just test backups for auditors anymore. We build systems that must survive when the power grid is hit, when the data center runs on diesel, and when the team is coding from shelters.

The biggest lesson for the global cybersecurity community?
Fragility is a choice.

We learned that secure architecture isn't about building unbreachable walls. It's about how fast you can stand back up when the walls shake.

To my fellow Ukrainians: We stand. We build. We defend.
To the global community: Don't wait for a crisis to test if your BCP actually works.

The photo date Feb 25th, the second day of full-scale war. My wife and I are relocating to Tuskavets.

Thank you, Creatio and Katherine Kostereva, for making it possible.

Is your resilience tested? 🤔

Ours is tested, every day.

AI is coming everywhere, and Cybersecurity is not an exception.

Kyiv International Cyber Resilience Forum was a blast. I have not yet seen so many cyber people in one place. This is one of the biggest Cybersecurity events in Ukraine to date.

The amount and intensity of the networking was unbelievable. Since I came to the forum at 11:00 a.m. I could not attend any of the panels or stages for the whole 2-2.5 hours, purely because of the number of people I knew and wanted to talk to.

The discussions just kept going, and I loved it.

The networking was clearly the main feature of the event. The people from Ukraine's Gov Cyberdefence, Startups, European Gov representatives, and Global startups.

The event was a "Cybersecurity Networking Academy Award" winner.

👇 What were the key topics for me?

1. AI is coming everywhere, and Cybersecurity is not an exception.

Hackers and Red teams using AI to find bugs, Defenders and Cybersecurity vendors using AI to defend. If you or your company are not using AI to find bugs or defend against them, you will become outdated and replaced very soon.

2. Cybersecurity community is growing day by day.

The demand for cybersecurity professionals is at all times high; companies that haven't done cyber before, such as SHERIFF, are now entering the market to defend not only security but also cybersecurity, as this is an inseparable element of privacy and safety nowadays. The wars start with cyber reconnaissance. The power grids, hospitals, schools, and business got attacked in the cybersecurity field. It's easier to apply, and it is not a head-on conflict as in physical space; it is abused a lot.

3. People are the weakest link in your cybersecurity chain (as it always been).

Global companies and governments got hacked because someone installed some suspicious Chrome spyware that stole the password to a corporate or gov account. People click on phishing links, not even knowing what they are or that there are emails, links, and attachments that should never be opened. Educate, educate, and then repeat. Regular cybersecurity speaking corners and mini-courses are a must nowadays. It’s not just about your company’s privacy and security; it’s about your personal privacy and security, too.

4. Ukraine is outpacing Europe in cyberspace.

Cybersecurity companies and professionals from Ukraine are growing fast, and government agencies are strong and cyber-resilient. Ukraine is already outpacing Europe in the Cyberspace, and is catching up to the United States very quickly. The professionals from Ukraine are in demand, and the companies are ready to pay top dollar for their experience.

Have you been to the event? What stood out to you? 😑

@securediary

TV Show: Burnt out and happy 🔥

Julia: Vlad, tell me how your day passes?

Vlad: Nothing special, I wake up at 5 am, then I work till 12 pm on the first full-time, then from 12 pm till 8 pm on the second full-time, and after 8 pm, that is it. I rest.

Julia: Oh, finally, after 8 pm, you have rest?

Vlad: No, I mean after 8 pm, I have a quick part-time job, a couple of tasks done, and $100 in your pocket.

Resonates with you? 🙂

@securediary

Claude AI Finds 22 Firefox Flaws, AI-Written Multi-Stage Attack.

Just another week in cybersecurity 🤷‍♂️.

Anthropic reportedly uncovered 22 new Firefox vulnerabilities in partnership with Mozilla. A strong signal that AI is becoming a real force multiplier for defenders.

The flip side is... Pakistan-based threat groups, such as Transparent Tribe, are also using AI to accelerate malware development. Chaining scripts, loaders, and social engineering - all combined into multi-stage attacks that look increasingly efficient and scalable.

AI isn’t just about making people more productive. It now gives an edge to those who use it well.

🔥 Top News to keep an eye on:

→ AI-assisted malware campaigns targeting India
→ Iranian-linked activity using the new Dindoor backdoor against U.S. networks
→ China-linked telecom attacks in South America using TernDoor, PeerTime, and BruteEntry
→ Microsoft disclosed a ClickFix campaign using Windows Terminal

What caught your attention?
What do you want to hear about? 👇

@securediary

🤔 You know the OWASP Top 10, but vulnerabilities still keep appearing?

🪄 Then it might be time to look deeper — and learn how to design secure systems from the start at the Security Architecture in Practice workshop.

📅 When: May 6 and May 9
👉 Registration & details: https://bit.ly/3OTAMWl

Most security issues don't happen because of a single bug — they happen because of architectural decisions.

In this workshop, we'll walk through real attack scenarios targeting web and SaaS systems, and discuss how to move from reactive "patching vulnerabilities" to a Secure by Design approach.

🤫 Mentor: @stansecure
11+ years of experience in information security, from serving in the Armed Forces of Ukraine to building security processes for global SaaS products.

What you'll gain from the workshop:

→ understanding of how attackers exploit weaknesses in applications and infrastructure
→ Secure by Design architectural patterns
→ practical Threat Modeling skills for applications and infrastructure
→ an understanding of how to integrate security into SDLC
→ ability to recognize the architectural causes of vulnerabilities, not just their symptoms

Discounts:
-30% for full-time students
-100% for active military or veterans (upd: all spots are booked)

👉 Sign up here: https://bit.ly/3OTAMWl

@securediary

How one phishing email led to a $X,000 payment to a scammer.🤑

It started with a compromised email account at a partner company P.
An employee clicked a phishing link, and their mailbox was taken over.

Why? Their Office 365 environment had zero security controls:

🛑 No MFA
🛑 No session expiration
🛑 No monitoring or SIEM
🛑 No security awareness training

So the compromise went completely unnoticed.

The attacker quietly stayed inside the mailbox for months, reading emails and watching how payments were handled.

Eventually, they saw their chance.

The employee responsible for payments received an email from what looked like a regular vendor, company A.

The domain looked almost identical.
Just one letter different.

Inside the email was an invoice for $X,000 😐.

The employee was expecting an invoice around that time, so the payment went through.

Later, they contacted the real vendor.

That's when they found out, the vendor never sent the invoice 🫢.

The bank account wasn’t theirs.

The money was gone. 🤫

Have you ever seen stories like this? How it happened?
What did you do?

@securediary

Google Chrome zero-days. Linux CrackArmor Flaws.

Pretty packed week in cybersecurity. Let me review what matters. 👀

1. Google fixed two Chrome zero-days already being exploited in the wild.

2. Researchers just dropped "CrackArmor" - nine separate flaws in the Linux AppArmor module. If you rely on AppArmor to isolate your containers, unprivileged users can now bypass those protections and escalate straight to root. It completely breaks the container isolation model you probably rely on.

3. Law enforcement (InterPol) disrupted the SocksEscort proxy network after it had abused hundreds of thousands of routers and IoT devices.

4. Veeam customers were told to move quickly on newly disclosed Backup & Replication flaws.

5. Microsoft is tracking a new campaign (Storm-2561) using SEO poisoning to push fake VPN clients. Employees are Googling for secure remote access tools, clicking the top result, and handing their corporate credentials straight to threat actors.

Are you using Google Chrome after this? 🤔

@securediary

I have a little anniversary on my Telegram Blog! 👀

Join up & share with friends @securediary

🔁 Reposts are most welcome.📣

I'll be at the Fwdays AI Summit in Kyiv on April 25.

One talk I'm looking forward to is "Defence-in-Depth: How We Build Security for Diia.AI" by Volodymyr Holomb from the Diia team. AI security in a product used by millions of Ukrainians -- worth showing up for.

A few other talks that caught my eye:

• Kateryna Stetsiuk will take an honest look at why 90% of AI pilots fail to deliver business results.
• Vyacheslav Koldovskyy is covering Agentic Engineering for brownfield projects.

Event details: https://bit.ly/4kCEvDg

Who else is going? 🤔

@securediary

A security scanner got backdoored. Twice. In the same month.

Aqua Security's Trivy was compromised by a group calling themselves TeamPCP. It started with a stolen credential, then force-pushed malicious code to 75 out of 76 version tags of trivy-action on GitHub, and turned the scanner itself into a credential harvester.

SSH keys, cloud creds, Kubernetes tokens, npm tokens leaked.

The first breach happened earlier this month. Credentials were rotated, but the attackers returned on March 19, publishing backdoored binaries to GitHub Releases and Docker Hub.

Then it got worse.

Using the npm tokens stolen from the Trivy breach, the attacker deployed a self-spreading worm (CanisterWorm) across 47+ npm packages.

It uses a blockchain smart contract as its C2 (command and control), which makes it much harder to stop.

You can’t just take it down by taking down its hosting provider.

What a week.

Blockchain malware, huh? Are you surprised? 🤔

@securediary

Would you agree? 👀

Why yes or no?