CyberDefendHERs Berlin 2026 – Talk „Responsible AI in Practice"
There was a line at the women's bathroom.
At a tech conference. In 2026, that's still — unusual. If you're a woman in tech, you know: that line usually doesn't exist. There aren't enough women in the room to create one.
In April in Berlin, it did.
CyberDefendHERs Berlin, the first-ever event of the initiative founded by Lisa Lobmeyer, hosted at ESMT Berlin, created the space that's still missing in most technical rooms. Elsa Berg Bjärtun (COO at Allurity) shared the number that explains why: only 14.6% of Germany's cybersecurity workforce is female. That isn't just a diversity metric. That's missing perspectives. And perspective gaps become blind spots in threat models and cyber defense.
My talk, „AI in Practice: Lessons on Risk, Accountability, and Organizational Reality," made three points:
AI risk is a whole-system discipline. As a systems engineer in the space industry, I learned that a satellite usually doesn't fail because one component fails. It fails because of how components interact under conditions nobody fully anticipated. AI is no different. The attack surface is a system, not a single vulnerability. Today's threats aren't static; with agentic AI, they evolve rapidly. Residual risk isn't something you leave undocumented — it's a conscious, deliberate decision.
Security, AI engineering, legal, and ethics have to be structurally in the same room. A single agent's risk profile touches training data integrity, regulatory exposure, security architecture, and organizational accountability all at once. No single domain owns that. Governance frameworks that leave any of these disciplines out — or that leave them coordinating through email — don't hold up in production.
The gap between what AI teams build and what security teams can evaluate is widening. Collective defense is the only response that scales. A vulnerability found in one organization is a warning signal for every organization running comparable systems.
What made the day different wasn't just who showed up. It was how many angles the same problem — resilience and cybersecurity in the AI age — was approached from: OT and critical infrastructure. Incident response in healthcare. Software supply chain risks in military and policy contexts. Threat intelligence through geopolitics and language. Deep tech innovation. Digital sovereignty.
Thank you to Lisa Lobmeyer for founding CyberDefendHERs, to Elsa Berg Bjärtun and the Allurity team for backing it, and to every speaker and participant who made the day what it was.
Where in your organization do AI and cybersecurity actually meet: in the org chart, or only in email chains?