Why T&S needs a Wikipedia-thon

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.

Last week, I guest taught a T&S class at Cornell Tech, where I showed how even simple-seeming policies can get extremely complex very quickly. It was rewarding to see the students grapple with tradeoffs and want to find the 'right' answer, only to find that every option had uncomfortable knock-on effects.

If I could, I'd make it mandatory to have anyone who talks about T&S issues go through a similar exercise, because you quickly learn how difficult the work really is. Unfortunately, much of the discourse around T&S these days is written by people who don't fully understand it, or don't even believe in it, which is a real problem. That's the focus of T&S Insider for the next few weeks.

Today's edition provides a real, tangible way to address this thorny issue. And if you're inspired to discuss further — not to mention meet other EiM readers — join Ben and me next week (Monday 28th September) for an hour-long chat. Spaces are limited.

Here we go! — Alice

Wikipedia editors, unite!

If you ask a regular person on the street about Trust & Safety, or how content moderation works, or about "the people who decide what they can say online", the answer — which may be positive but will more likely be negative — will be derived from one of the following groups of messengers: journalists, academics, civil society, filmmakers, lawmakers or regulators.

The challenge, as we all know, is that T&S is an extremely complex field, and the tradeoffs and restraints that we have to work under are hard to understand unless you’ve done the work yourself. Ben said it himself just recently:

"I think there isn't a shared view right now of how trust and safety tells its story. I know that people think trust and safety is done best when it's hidden, when it's in the shadows, when people aren't seen. But I know from the media industry for example, that actually that's not always the best way for engendering trust."

- Ben Whitelaw, Safe Space Podcast (clip here: LinkedIn, Safer built by Thorn)

Now, I don't believe we can leave the telling of our story to anyone outside of the field, which is a large reason why I write for EiM. And while other practitioners have written about T&S in Tech Policy Press, TSPA’s curriculum, the Integrity Institute, and elsewhere, those sources — they won't mind me saying — mainly talk to the converted.

But, there’s one source which is larger than all the above combined and which is woefully behind in its understanding of what T&S is: Wikipedia. 

Why Wikipedia?

You don't need me to tell you that Wikipedia is the largest and most read reference work in history. It catalogs everyone and everything of note, especially internet topics. And it’s where a regulator, a journalist on deadline, a new hire and an AI assistant all start.

It's also one source that is very hard to game. Attribution, independent sourcing and balance against critics are mandatory. Editors only accept updates to pages with proper citations and, when that doesn't happen, it often becomes a news story in its own right. Naturally, Wikipedia has its own page dedicated to "political editing incidents".

However, if you relied solely on Wikipedia to understand the vast, complex, and extremely nuanced world of Trust & Safety, you’d come up extremely short.

Red links and redirects

My friend and former colleague Juliet Shen turned me on to this. She noticed that the T&S page was particularly bad so she spent some of her spare time editing it to be more accurate and relevant. Even after the heroic work she’s done over the last year to update articles (page edits here), the state of coverage of T&S is still shocking to me: 

• Content moderation is focused almost completely on content moderators and the harms of moderation. It doesn’t give details on automation or risk or policy frameworks. The navigation template at the bottom of the page is for censorship and websites.
• The page for Del Harvey, who ran Twitter's T&S function for a decade, redirects to Perverted-Justice and makes no mention of her contribution to T&S. There’s no page for Dave Willner, Charlotte Willner, or other notable T&S trailblazers.
• The Integrity Institute has never been cited on Wikipedia, despite years of published research by T&S professionals. 
• Content policy is a red link. Platform policy redirects to Online Community.
• Also missing are Santa Clara Principles, Trusted flagger (a defined role under DSA Article 22), Platform governance, and TSPA and Trust and Safety Professional Association.
• Wikipedia files content moderation under Outsourcing and Reputation management, which is baffling.
• The Trust and Safety page is rated importance "Unknown" despite all the additions and changes that Juliet has made to it recently. Category:Trust and safety does not exist. 

And there’s more. 

Frankly, it's hard not to conclude that the major public record of T&S — accessed almost 400 times a month and more than 50,000 times over the last decade — has largely been written by its critics and its dissidents, rather than anyone with knowledge of the industry.

What we can do about it

A Wikipedia page can change quickly, if we all work together.

Remember that the Trust and safety article was only created in March 2023 and sat as a tiny page for two and a half years before Juliet took it upon herself to. The edit log before that point is mostly vendors adding themselves to a "Companies" list and volunteers reverting them. "Removed Google promotional content." "Removed self promoting plug." etc.

Any of us can do the same thing. If you have a little bit of time, here are some examples:

• Ten minutes: Add a citation. Lawfare and Tech Policy Press are already accepted as sources. Integrity Institute research isn’t cited at all, but should be. 
• An hour: Update a thin page, like Shadow banning, which describes the downsides but not why platforms do it. 
• A weekend: Write about a missing concept or institution, such as Santa Clara Principles, or Trusted flagger.

Over time, we can also:

• Create and populate Category:Trust and safety or Category:Trust and safety professionals (For comparison, Category:Artificial intelligence researchers has 303 members, meaning we're 303 behind).
• Devise a WikiProject with an assessment scale.
• Track the uptick in views of the Trust and Safety page as a result of edits we've made.

Getting started

If you've read until this point, I'm going to assume you're interested in helping. Thank you for doing so.

I've created a spreadsheet to get us started, which, in the spirit of Wikipedia, can be viewed and edited by anyone. Pick an edit and update the status column once you're done.

If you've never edited on Wikipedia before, don't worry:

1. Make an account and say who you work for (here’s mine that I just created).
2. Make a few small suggested/ easy edits before tackling anything contested.
3. Pro tip: Propose edits on the talk page of an article before restructuring or rewriting completely. 

I’ll close with a quote from Ben again: 

"Do we want to remain fairly silent, fairly behind closed doors, which is one way which is completely valid, or do we want to slowly and responsibly go out and have a broader discussion in order to engender that trust?"

The most trustworthy source is Wikipedia. Let’s make it a true reflection of our field. 

Also worth reading

EU KIDS Act to restrict social media platforms' access to children in the EU (European Commission)
Why? The Commission's proposal could become the most far-reaching child online safety regime yet. It bars under-13s from social media and reverses the burden of proof, so platforms must show their products are safe by design.

What if social media isn't hurting kids? (The Verge)
Why? An interview with Peter Gray, whose book Restoring Childhood challenges the Haidt-led case for restricting youth social media, arguing that large-scale research finds no significant effect on teen mental health.

AI-Generated CSAM after Anderegg: What Changes (And Doesn't) on Product Roadmaps (Quire, by Vys)
Why? The Seventh Circuit's ruling is narrow, covering only private possession of obscene AI images depicting no real child, and it doesn't change platforms' reporting duties. The piece corrects the common misreadings and lists what teams should revisit, including possession-based policy language, detection gaps for novel content, and negative prompts in red-teaming.

Scam spotting with ChatGPT (OpenAI)
Why? OpenAI shares data on how people use ChatGPT to check whether something is a scam. Read it for a rare platform-side look at scam-checking behavior at scale.

Detecting and Countering Misuse of AI: September 2026 (Anthropic)
Why? Covers disrupted operations across cyber, influence operations, scams and fraud, and other harm categories from December 2025 to August 2026. Its main finding is that agentic AI has collapsed the sophistication gap, letting small groups run campaigns at machine speed that once needed state resources.