Varshith H Anilkumar

Speaking

Berlin Science Week 2026 · Inspiration Stage · 7 November 2026

Berlin Science Week 2026 · Inspiration Stage

Can AI Be Intimate Without Being Invasive?

Session
Inspiration Stage: Between the Machine & Me
When
7 November 2026, 12:00–13:30
Where
Angie Stardust Foyer, Festival Centre at Haus der Kulturen der Welt (HKW), Berlin
Speaking as
Independent Researcher
Festival theme
In Touch

We confide in AI the way we confide in a friend, but its memory belongs to a company. This talk asks whether chips modeled on the brain could change that.

AI is becoming a constant companion. We confide in it, and it is learning how we feel, which is the most intimate data there is. What protects that today is a promise, a privacy policy, or the assurance that your data never leaves your device, and neither survives a hard fact: trained models themselves leak, and what a model has learned about you can be pulled back out of it. This 12-minute talk asks a different question: could privacy be something a machine is built to be incapable of breaking, rather than a rule we hope it follows? Along the way, it takes an honest look at one contested lead, chips modeled on the brain, where published research first found the usual attacks losing much of their power, and newer work then showed that the protection is not guaranteed. There is no finished answer yet, which is exactly why the question matters now. No technical background needed: you will leave with one question to test against every device you own. Can AI be intimate without being invasive?

Topics

Each of these maps to work I am actually doing rather than a talk track assembled for the stage.

Security and trust in computing systems

Information security and data governance as they are actually run inside a regulated European health-tech company: ISMS build-out, special-category data, vendor due diligence, and what trust has to mean when it is engineered rather than promised.

AI governance and the EU AI Act in practice

Risk classification, documentation and human oversight as engineering work rather than paperwork: what the Act asks of product and engineering teams, and where the obligations bite in practice.

Neuromorphic and frontier compute

Could privacy be something a machine is built to be incapable of breaking, rather than a rule we hope it follows? There is no finished answer yet, which is exactly why the question matters now.

What production AI actually takes

A decade of shipping AI and ML systems at scale, across payments, commerce, IoT and digital health, and the distance between a demo and a system that survives real users, real data and real compliance requirements.

Booking

Send the event, the date, the audience and the format. I'll tell you honestly whether I'm the right person for it.

Enquire about a talk

For workshops and board sessions, see work with me.