Toronto AI Safety

Getting here: Enter the lobby at 100 University Ave (right next to St Andrew subway station), and message Giles Edkins on the meetup app or call him on 647-823-4865 to be let up to room 6H.

Sometimes we just like to break things!

Research into attacks against machine learning systems is vital. A system that might appear robust may fail in the presence of carefully crafted inputs or training data. Knowing the possible avenues of attack can help with defense.

We'll talk about different kinds of attack, some defenses, and which machine learning use cases are particularly important to make robust.

We welcome a variety of backgrounds, opinions and experience levels.

**IMPORTANT NOTE** There will be a film crew, working on a documentary about AI risk, filming today's meetup! If you show up, you'll be on camera and have to fill in a consent form. If you're interested in the topic but don't want to be filmed, we'll provide an option to join remotely over Google meet.

Getting here: Enter the lobby at 100 University Ave (right next to St Andrew subway station), and message Giles Edkins on the meetup app or call him on 647-823-4865 to be let up to room 6H.

Kathrin Gardhouse will be here to talk about the EU AI Act.

The EU has passed an important piece of legislation regulating the use and development of AI systems. But what exactly does this law say, and does it address the most pressing AI safety concerns - both immediate and longer term?

As always, there'll be plenty of time to interact and ask questions. Join for an interesting discussion on AI politics and governance as it manifests in the real world.

Training the most capable AI systems currently requires huge amounts of compute, usually in the form of GPUs specially optimized for this task. The trend of huge training runs is likely to continue for a while.

At the same time there is increasing interest in regulating AI, and a big question mark over the inherent safety of possible "human-level" AGI systems. This raises the possibility of intervening at the compute level - monitoring the largest training runs and the computers that they run on.

Is this a good idea? What measures could be put into place to balance the need for AI safety while preventing authoritarian overreach by governments? Mario will lead a presentation and chat!

All levels of knowledge and experience are welcome.

Join a friendly and intelligent group as we discuss the future of AI and Machine Learning and how to help ensure the dramatic changes will go well.

Expect debate, presentations, some technical content and some practical (programming) workshops. Each week we'll let you know what to expect as the different events will each have a different flavour.

We welcome a variety of backgrounds, opinions and experience levels.