AI Safety Meetup #1
AI is getting smarter and more autonomous every month. Let's talk about how to make it go well.
AI Safety Meetup is a new monthly series at Kolektyw3: a casual conversation with no forced slides and no sales pitches. We sit down together and talk about building and deploying AI responsibly.
What we talk about (for example):
alignment: how do we know a model does what we want, and isn't just looking like it does?
interpretability: what's actually going on inside the network?
evals and red teaming for dangerous capabilities
autonomous agents: how much freedom is too much?
regulation and governance: the EU AI Act, lab safety policies, standards
how to get into AI safety work, including from Poland
Who it's for: researchers, engineers, founders, policy people, and anyone who finds this topic fascinating or a little worrying. Any level of experience is welcome; curiosity is the only requirement.
Format: a short intro to the topic, then open discussion and networking. Have a paper, project, or idea you'd like to share in 5 minutes? Mention it when you register.
When: once a month on a Wednesday (on the Wednesdays without Vibecoder Wednesday).
Hosted by Konrad Pabiańczyk & Kolektyw3