Are we entering Trust and Safety’s builder era?

I'm Alice Hunsberger. Trust & Safety Insider is my weekly rundown on the topics, industry trends and workplace strategies that trust and safety professionals need to know about to do their job. This week, I'm thinking about:

It was a blast seeing so many T&S Insider readers in San Francisco last week; thank you for all the kind words and to the various speakers that mentioned EiM up on stage. I'm still recovering from what was a brilliant TrustCon.

As with 2024 and last year, I'm using this week's edition to reflect on what I heard and saw at the conference. I hope it's useful for those who couldn't make it; if you want to pick up on any of the threads in today's newsletter, get in touch.

Thanks for reading, here we go! — Alice

What TrustCon's good mood couldn't show me

I went to this year’s TrustCon expecting a subdued atmosphere. The last year in trust and safety has been full of layoffs, restructures and uncertainty. However, more than a thousand people attended this year's conference and, in all of the many hallway conversations I had, nobody brought up burnout or wanting to leave the industry. Despite the chaos of the last few years, people seemed energised and mostly in a good mood.

Obviously, some of that is down to the type of person who typically comes to an industry conference. Not to state the obvious but the people who made it to San Francisco last week are a) people who still have jobs and b) people whose managers who have budgets and are willing to send them. Even if they might not feel like it, they are the lucky ones of the T&S industry. Others didn’t make the trip either because they couldn’t or for fear of the US administration might do something to prevent them from travelling safely.

That split is consistent with the wider tech industry. A recent survey by Noam Segal and Lenny Rachitsky (of Lenny’s Newsletter) covering around 6,000 tech workers, found one half buoyed by and benefiting from AI, and the other part burned out and worried about the future. The happier half skews toward smaller, better-run teams, whose employees are the people most able to travel and least likely to have just watched their own team get cut. Sounds familiar, doesn’t it?

Build, baby, build

I spent most of my time at TrustCon with this lucky, optimistic group of T&S professionals. And it was increasingly clear why they felt this way. 

For the first time, using LLMs and self-coded tools, Trust & Safety teams have been able to build the things they used to file engineering tickets for. Policy managers can run policy research in hours rather than weeks. Enforcement specialists can automate the most repetitive and most damaging review decisions. Almost anyone can point an agent at a database and ask it questions rather than wait for a data analyst or engineer to pull it for them. For a field that's spent its whole history under-resourced and working reactively, that feels like a different job, and a better one. 

That urge to build and to do came through strongly in the TrustCon schedule. There were sessions on: How to build the classifier; how to structure a policy so a model can enforce it; how to evaluate an agent; how to wire detection into the pipeline and more. Some of this was a result of the rooms I picked, but TSPA also made a point of having more sessions on AI and practical building than ever before. 

My big takeaway was that the T&S capabilities that used to require significant budget and extensive sign-off are getting cheaper and more accessible, with many teams and vendors also sharing free resources openly. That means the gap between what a well-funded team can do and what a small team can do is narrowing.

Introducing Safe for Work? — all about T&S jobs

This new T&S Insider series will look at the T&S job market in 2025 and unpack what it means for people who want in (or out)

Everything in Moderation*Ben Whitelaw

The enduring need for T&S leaders

The builder energy I’ve described depends on two things that strained teams don't always have:

1. Leadership that treats AI as a way to produce higher-quality work, rather than simply cheaper work
2. Enough senior people to architect the systems in the first place

If the leaders in your company think better means only doing more and doing it faster, AI tools and LLMs might make your job harder, because they raise the ceiling on how much you're expected to produce. For small teams that are already stretched, that’s not going to end well.

As for the architects problem, that could be about to get worse. Many teams have fewer juniors who would have gone on to become the next architects. Not only does that mean getting into the industry is harder — which I’ve written about — but it concentrates the ability to use these tools well in the people who already have the skills and knowledge. 

In short, it doesn’t matter how accessible the tools are if the conditions that allow them to pay off — that is, supportive leadership and sufficient senior expertise — are only enjoyed by a small set of teams at companies that already had those conditions.

Despite all of that, I came home feeling excited about T&S work, and I think many other people did as well. But I’m conscious that I spent those three days almost entirely with the people for whom this moment is going well. The ones for whom it isn't just weren't there to say so.