Goran

Goran

Croatia, EU · Information Security Lead · Secro Inc

University of Zagreb (Sveučilište u Zagrebu)

17 followers

About

Goran Pizent is an Information Security leader with over 20 years of experience in CISO-level strategy, DevSecOps engineering, regulatory compliance, and blockchain/fintech security. He has a proven track record of owning SOC 2 Type 2 and ISO 27001 programs end-to-end and possesses deep hands-on expertise in cloud-native infrastructure on AWS and GCP, Kubernetes, and software supply chain security (SLSA). Goran is experienced in presenting risk posture and compliance status to executive leadership and boards in regulated financial environments.

Skills

  • ISO 27001
  • SOC 2 Type 2
  • Risk Register & Risk Treatment
  • Board-level Reporting
  • Incident Response (SIRT Lead)
  • Business Continuity / DR
  • AWS
  • GCP
  • Kubernetes
  • SAST

32 skills in total

Experience

  • Information Security Lead

    Secro Inc

    Aug 2024 — Present

    Remote / EU

  • Senior DevSecOps Engineer

    Parity Technologies

    May 2023 — Aug 2024

    Remote

  • Senior DevSecOps Engineer

    ThreatMark

    Sep 2022 — May 2023

    Remote

13 experiences in total

Education

  • University of Zagreb (Sveučilište u Zagrebu)

    MSc, Computer Science

    Jan 2001

Recent posts

Agentic pentesting tools are no longer confined to security labs.

CrowdStrike found ARTEX configurations, Claude Code histories and model prompts on infrastructure connected to attacks against South Korean financial organisations.

The interesting part for me is the workflow. The actor combined an autonomous testing tool, several models, proxy infrastructure and conventional offensive techniques.

I would start hunting for the automation around the intrusion, not only the final payload.

Can your detections distinguish a developer running an AI security tool from an attacker doing the same thing?

https://www.crowdstrike.com/en-us/blog/unknown-threat-actor-uses-artex-to-target-south-korean-finance/

#ThreatDetection #AISecurity #FinancialSecurity

View post

Is this the end? GLM-5.3 changes the scenery.

Developing a working cyber exploit may be shifting from a specialist task that takes days or weeks to something that costs about $20 and needs around 20 minutes of human attention.

That is what caught my attention in Anthropic’s research on GLM-5.3.

In one of their tests, GLM-5.3-Flash used publicly available vulnerability information to build a working ARM64 exploit chain. Anthropic also tested GLM-5.3 on previously unknown browser vulnerabilities and found that it could discover vulnerabilities and combine them into an exploit capable of reading files from a victim’s machine.

For me, the interesting part is not that AI can write exploit code. We already knew that was coming.

It is the reduction in effort.

Time, specialist knowledge and cost have always limited how many people can develop advanced exploits. If models keep reducing those barriers, the number of people capable of carrying out fairly sophisticated attacks changes dramatically.

That also changes vulnerability management.

The time between a vulnerability becoming public and somebody turning it into a usable exploit could become much shorter. A patching process designed around days or weeks may increasingly be competing with AI agents working in minutes.

There is another side to this. Security teams have access to these capabilities too.

I would expect more vulnerability research, automated security testing and eventually AI agents actively looking for weaknesses in our own environments before someone else does.

So I do not think this is the end.

But I do think GLM-5.3 is another sign that the assumptions we built vulnerability management and offensive security around are starting to change.

And probably faster than many organisations expect.

https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities

#CyberSecurity #AISecurity #ArtificialIntelligence #VulnerabilityManagement #DevSecOps

View post

If an AI agent pushes code, whose name appears in your audit trail?

NIST’s latest DevSecOps update includes plans to test how agents are identified, authenticated and authorized within the SDLC.

That’s the part I’d want to see working. When an agent requests a secret or starts a build, I want to trace the action to its delegated task and the permission used.

Can your pipeline distinguish the agent’s actions from the developer who launched it?

https://www.nist.gov/news-events/news/2026/09/new-nist-nccoe-resources-devsecops-and-october-28-webinar-agentic-ai

#DevSecOps #AISecurity #IdentitySecurity #NIST

View post

I asked for a sales report. Why does the AI agent have access to payroll?

I can see a reason for the agent to read CRM records and the relevant contracts. I can’t see a reason for it to browse payroll, even though I happen to have access to both.

This is the part of Proofpoint’s recent announcement that interests me: connecting what an agent is doing with the data it can reach. The announced capabilities are expected by year-end, so I would like to test them before drawing conclusions.

In a review, I would ask someone to show me the actual permissions behind the demo. If the agent uses my session, how do we restrict it to the work delegated? What happens when it decides it needs another source?

I expect some friction here. An agent will stop and ask for access, and some tasks will definitely take longer.

I am comfortable with that.

Giving it my full account access to avoid an approval step is a trade-off I still cant justify.

How about you?

View post

View Goran's profile on Cyberr