Prioritise writing and sharing

How to write well

How to share your work

Get it working

Figure out what needs to be true

Deciding what to test first

Using your results

Evaluation criteria

Submit your project

Next steps: Programs

You can make AI systems safer

Getting started

Project ideas

Scope it down

Get funding for your project

We’ll be reviewing your project based on three dimensions. Here’s what good looks like:

**1) Clarity and presentation**
Your write-up has a clear question or goal, and someone unfamiliar with your project could pick it up and understand what you did, what you found, and why it matters. The best projects are genuinely easy and engaging to read.

**2) Relevance to AI safety**
You explicitly explain how your work connects to the safety of advanced AI systems — not just AI in general. The strongest projects explain how they fill a gap in existing work or have a clear theory of change. Projects with negative results ("I tried this approach and here's why it didn't work") are valuable too, as long as you explain what that tells us.

**3) Project quality**
The project demonstrates care in execution — clean implementation, sensible choices, clear reasoning. It doesn't need to be novel. The best projects show strong judgment about what matters, include honest limitations, and suggest concrete next steps. The write-up should make someone working in this area want to see the work continued.

You can expect to receive feedback within 2 weeks from the submission deadline.


If you're on a facilitated version of the course, submit your project here to get feedback. 

Congratulations on the hard work you put in 🎉 

We will feature the best projects on [our website](https://blog.bluedot.org/s/projects). We may also share them with other orgs with relevant jobs or opportunities. 

We strongly encourage you to pin your project on GitHub and make at least one post on X or LinkedIn about it.


Technical AI Safety Project

Evaluation criteria