Is it time to unite T&S and AI ethics?

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.

Today I'm thinking about:

• How AI Ethics and Trust & Safety teams can collaborate and innovate for everyone's benefit
• How US federal workers can make the move to Trust & Safety

As always, get in touch if you'd like your questions answered or just want to share your feedback about today's edition.

Here we go! — Alice

Today’s edition is in partnership with Safer by Thorn, a purpose-built solution for detection of online sexual harms against children

At Thorn, we are committed to pushing the boundaries of innovation in the trust and safety space, developing purpose-built solutions to protect children from harm in the digital age.

We’re proud and grateful to be selected as one of Everest Group’s “Content Moderation Technology Trailblazers.” Everest’s recent report celebrates and calls attention to top tech startups creating buzz in the industry, and our inclusion on this short list reflects the impact and innovation we’ve long prioritised.

T&S and AI ethics is not an ‘either or’ choice

If the last two years have been defined by two major trends, they would be AI’s rapid expansion and the erosion of trust in content moderation.

AI has rapidly become central to online platforms, powering everything from creative content to routine task automation — but it has also introduced new risks and harms.

At the same time, mass layoffs in Trust & Safety and widespread scepticism of content moderation and enforcement have sent a mixed message: we’re going to invest in AI, but not in responding to its risks. 

The tension between these two industry-wide shifts means we might need to drastically rethink things. Let me explain why.

...

Trust & Safety teams were originally set up to handle human-centric issues and ensure community safety through guidelines and enforcement. This work continues today, but has expanded to include reviewing AI-generated content and using AI enforcement algorithms.  As a result, academic institutions are developing research capacities and teaching resources to cater for the demand. What started as a practical discipline is gradually becoming the focus of academia.

In contrast to T&S, the discipline of AI ethics started life in academia and research, and is now finding a home at platforms focusing on algorithmic fairness, transparency, and responsible deployment of technology. Not only that but, according to a recent IBM report, 75% of executives view AI ethics as a competitive differentiation, and over 85% of surveyed consumers, citizens, and employees value AI ethics. This is an academic discipline slowly working its way into corporations around the world.

As you can see, there are similarities and differences between T&S and AI ethics. Both disciplines put company values into action by designing systems, guardrails, and governance structures. They are also both used to build trust with users. But in practice, T&S and Ethical AI teams often work separately and in silos, mostly to the different origin stories of each type of work.

In the past, this made sense. Human-made harms and AI safety were distinct business areas. However, the work is now completely interconnected and fundamental to almost every platform’s core business strategy. 

Risks don’t come in neat, compartmentalised packages and we can no longer separate “AI” and “human” — it will always be both, together. Humans use AI to accelerate the creation and distribution of illicit content. T&S teams are using AI to detect and remove this content and find anomalous behaviour faster. AI systems have their own potential for bias and platforms are encouraging users to generate AI-created content, making it difficult to detect what is AI and what isn’t.

In short, it is all intertwined. Algorithms shape people’s experiences of the platform, and people’s use of AI shape the algorithms.

As the first AI Safety Report says, the time is now to focus on these issues:

“AI does not happen to us: choices made by people determine its future. The future of general-purpose AI technology is uncertain, with a wide range of trajectories appearing to be possible even in the near future, including both very positive and very negative outcomes. This uncertainty can evoke fatalism and make AI appear as something that happens to us. But it will be the decisions of societies and governments on how to navigate this uncertainty that determine which path we will take.”

Better together?

Increasingly, I’d like to see T&S and AI ethics come together as one discipline. Rather than be treated as separate functions with distinct teams, there’s scope to bring them together, encourage innovation and enable platforms to address issues and solutions holistically. 

Of course, there may be hurdles in combining these disciplines:

• There will be natural cultural differences between AI ethics and T&S teams due to their separate evolution
• Metrics, measurement, standards, and jargon would need re-evaluation
• Specialised expertise must be respected to maintain each team’s value

However, it seems riskier to continue acting as though we can separate AI vs. human, and ethics vs. safety. I don't think we can.

Some organisations agree and have made steps to bring them closer together. One company explicitly pairing AI ethics and T&S is Salesforce, with their Office of Ethical and Humane Use. Rumman Chowdhury, a well-known AI Ethicist, also gave a keynote speech at TrustCon last year, in which she talked about using AI to innovate and make user experience safer.

Yet these are mostly one-offs. For now, I mostly see AI and T&S work siloed and separate, and increasingly being positioned as ‘either or’ within platforms.

More than a cost centre

So what could that look like? 

I’m going to take us back to that IBM report that I mentioned earlier and a quote that may sound surprisingly familiar to those working in Trust & Safety:

“Traditionally, investments are justified by calculating ROI in financial terms alone. AI ethics investments are more challenging to evaluate, providing both tangible and intangible benefits as well as helping build longer-term capabilities.

“Our work has to not just contribute to the mission of the organization— it also has to contribute to the profit margin of the organization,” notes Reggie Townsend, VP of the Data Ethics Practice at SAS. “Otherwise, it comes across as a charity, and charity doesn’t get funded for very long.””

The paper outlines ways to justify investments in AI Ethics that would also work with broader joint initiatives including Trust & Safety. They begin with two pretty well-known justifications for safety-oriented work: traditional return on investment calculations (I warn against falling into that trap here), and measuring reputational impact and erosion of trust (a strong argument, but one that requires a leap of faith before the long-term benefits weigh out). The third suggestion is what they call “capabilities”, encouraging teams to use their capabilities to create broader value throughout an organisation.

It's this third one that I'm most interested in. Considering T&S beyond policy and enforcement expands our capabilities. Combined with AI ethics and governance work, there is scope to become more than just a cost centre.

A bridge between customer and company

Some ideas where I think AI ethics and T&S working closer together would be helpful include:

• Using behavioural frameworks for risk mitigation to identify prosocial behaviours and super-users, helping to drive positive platform experiences.
• Governance and ethics principles can build employee loyalty and alignment with company values.
• Feedback and transparency loops required by safety regulators can be used for other kinds of user communication and trust building.

There are many more possibilities to explore, especially if teams collaborate and are given the freedom to be proactive and innovative, rather than purely reactive and operational (as is often the case for T&S, unfortunately).

Trust & Safety and AI ethics work should be seen as a bridge between customer experiences and company values, and as leading experts in the many messy and beautiful ways that AI and humans can interact with each other. If T&S and AI ethics teams can work with each other and learn from each other more, then maybe they’ll have the impact that is so sorely needed at digital platforms today.

Resources for US government workers

We've all seen what's being called the ‘Rapid Unscheduled Disassembly’ of the United States Government'. If you know someone who is affected or is looking for a change of scene, here are some resources to share:

• Hinge's Jeff Dunn writes about how government workers can transition into the T&S Industry.
• Trust and Safety Professional Assocation's from government to T&S webinar.
• All my career links in one place.
• Theo Skeadas shares her perspective on LinkedIn, as a former Twitter employee and Government consultant (also check out her links in her first comment).
• Ask A Manager answers questions from federal workers who are currently under attack. (Read the comments, too)

Also worth Reading

Burnout in Trust & Safety: The Evolution of Wellness and Resilience in Integrity work (Trust in Tech Podcast)
Why: I recorded this a while ago, but couldn't get it together enough to edit it. Now it seems like the timing is perfect!

What Comes after DEI (HBR)
Why: Some interesting thoughts on how to move forward on DEI initiatives while improving outcomes.

Tracing the Speech Regulation Patterns of 2025 (Tech Policy Press)
Why: A lot has happened so far!