Now that you broadly understand what it takes to make AI systems safe, we can think through the threat pathways and how we might build defences.


Building defences

What can _you_ do to start contributing?


Make a plan

How can you contribute to making AI safer?


Why is building safe AI so technically difficult?


How do we teach AI to behave the way we want?


What's actually happening inside these black boxes?


How do we know if the AI is actually safe?


What happens when training and detection fail?
