TLDR:

The field of AI safety is bottlenecked on talent. Running recruitment processes is expensive and time-consuming. Freelancers are overlooked. Hiring freelancers can provide a way to quickly and cheaply test a person's fit within an org, and vice versa. Plus, real work gets completed, and freelancers get both compensation and a portfolio piece.

Last week, I wrote about my experience of transitioning into AI safety as a generalist in 2026. Immediately after hitting post, I checked my email and had my first offer for paid work in the field. Reading all of the comments from other generalists sharing their experiences on my last post, I felt very lucky to have some return on the time and effort I have invested into this transition.

One thing I keep thinking is that if I were looking for a full-time role, I would probably not have received an offer at this point. It is much less costly for an organisation to pay for a couple of days of work from a freelancer than it is to hire for a full-time role. If the person isn't a good fit for the org, some mediocre work is created, but the org wouldn't be at a huge loss. A couple of my conversations at EAG London 2026 went something like:

• Them: "I'm looking for someone to do x work"
• Me: "Great, I do x work"
• Them: "Let's try a small project together and see how it goes?"
• Me: "Sounds good!"

In the best case scenario of an org hiring a freelancer, the project goes well, the freelancer gets paid, and real work gets completed that the org wouldn't have been able to complete otherwise, i.e. counterfactual impact is created. Even better, this could lead to ongoing freelance work or a full-time role, either with the original org or another org to which they were able to refer the freelancer.

In London two weeks ago, I spoke to Maja Webster Nenadov, who founded Freelancing for Good. She was very surprised to hear from multiple EA orgs that they had not considered hiring freelancers to plug the gap between the work they wanted to do and their current capacity. When I shared insights from the reading I had done about the need for generalists in AI safety, we agreed that solving this problem could help with the talent bottleneck.

I'm exploring the potential for a project/org that will help AI safety orgs to hire those looking for work in the field as freelancers/contractors. I have a lot of unknowns to figure out and have begun to schedule calls with many of the generalists who commented on my post last week in order to figure these out. I also want to speak to hiring managers or anyone else who has a good sense of why orgs aren't hiring more freelancers.

Some of the questions I want to get answered by generalists:

• Could you do freelance work if you wanted to? If not, what are the barriers?
• Does the country you live in make it easy to try out freelancing?
• Would you be open to freelance work? Why/why not?
• Could your work (e.g. in operations, communications) be projectised? If not, why not?
• Does your role need a lot of org-specific context?
• What worries would you have in taking on freelance work?

Some of the questions I want to get answered by hiring managers:

• Has your org considered using freelancers? Why/why not? How has it gone so far?
• What work that your organisation needs is most suited to being handled by a freelancer?
• In general, what types of roles that are currently in demand within AI safety (communications, operations, etc.) are best suited for a freelancer to take on?
• What worries would you have in contracting a freelancer?
• Do freelancers need to signal alignment and/or understanding of the field in order to work with your org?

If you believe you are well-positioned to answer any of these questions, please feel free to leave a comment, message me privately, or schedule a meeting where we can talk more freely.

Further thoughts:

This project idea is still very much in its infancy, and the form that any solution may take is still unknown. I wanted to share some ideas for what it could look like anyway:

• A curated directory or roster of vetted freelancers with AI safety alignment, searchable by skill type (design, ops, comms, engineering) - adjacent to Freelancing for Good's directory
• A pilot program where a small number of orgs each commit to one scoped freelance project, with structured feedback afterward
• A Slack channel or lightweight community where orgs post scoped work and freelancers can express interest
• A standard "freelance starter pack" for AI safety orgs, including contract templates, onboarding checklists, project scoping guides

Adjacent project idea:

• Outreach to freelancers who are seeing their work be destabilised due to AI, i.e. those being replaced/scared of being replaced by AI - offering them a track into a well-funded field could be very impactful