What the T&S community predicts for 2026

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

This week, I'm off to the West Coast for a work onsite, and getting ready to have a lot of strategy conversations. I hate New Year's predictions, but I do appreciate the opportunity to take a longer view than the current news agenda/day-to-day chaos allows. In the newsletter today, we share what T&S experts believe will happen in 2026.

Get in touch if you'd like your questions answered or want to share what you like/dislike/want to see in the future. Here we go! — Alice

We asked, you predicted

Is mid-January too late for annual predictions? Maybe. But last year's T&S Insider 2025 look-ahead — courtesy of the EiM subscribers and practitioners on LinkedIn — was bang on the money, so it makes sense to do it again.

What we didn’t predict was quite how fast — or how chaotically — it would all unfold. But, frankly, who did?

For example, we saw the wave of regulatory impact coming, but we didn't predict Trump's executive order on "federal censorship" or the State Department denying H-1B visas to foreign T&S workers. We also imagined the job market evolving, but changes happened faster and more dramatically than any of us expected.

This year, as well as canvassing the wise T&S folks of LinkedIn, I've also reviewed all the 2026 prediction emails in my inbox and layered in the conversations I’ve been having with T&S leaders lately.

Here's the three themes that came out most strongly, each with ideas about how you can prepare if they come true. Because that, after all, is what matters.

Expanding beyond content and identity

If 2025 was about AI-generated content, 2026 will be the year of AI agent behaviour.

Ljubiša Velikić, VP of T&S at TELUS, put it bluntly: "AI to AI interactions [will become] a safety problem." When AI agents can independently book appointments, make purchases, or submit forms on behalf of users, the attack surface expands exponentially. Bad actors will deploy AI agents to distribute content at scale, evade detection, and adapt in real-time.

Similarly, platforms will be forced to expand their work around identity as digital IDs become easier to fake, but regulators are requiring more platforms to verify identity and age. This was a clear theme in the comments: Assaf Kipnis, founder of KTLYST Labs, predicted, "identity verification will start fading away as it becomes a tool of the past."

Velikić also predicted that "synthetic identity abuse explodes," while Payal Shah, a freelance AI/ML product leader, warned that "deepfakes and synthetic identity fraud are already exploding and will force a fundamental rethink of identity and risk."

How to prepare

• Invest in behavioural analysis, not just content detection — especially tools that can adapt automatically.
• Start red teaming now. Most detection systems are built around human behaviour; map where they’ll fail against autonomous agents.
• Don’t wait for regulators to define “safe” agentic AI. Connect with other T&S teams working on AI-to-AI interactions and co-develop best practices.
• Accept that document or selfie-based ID checks won’t scale. If age verification is required, implement it — but back it up with layered signals like payment history, social graphs, and device reputation that are harder to fake at scale.

Expanding roles and industries

Commenters on my LinkedIn post couldn’t agree on whether all of T&S was going to be replaced by AI or whether AI was going to supercharge their work, but no one denied the fact that AI will continue to change things. Kira Osborne, a consultant and former Australian eCommissioner lead, summed it up when she said, "We might just be swallowed whole but the very systems we're fighting to safeguard ourselves against! The irony....."

I believe that T&S workers of the future won't be making one-off decisions on individual pieces of content; they'll be managing hundreds of AI agents, becoming experts at data analysis, and having their decisions amplified and enforced at scale instantly. I’m just not sure how fast that will happen.

I'm a little more sure on the fact that 2026 will be the year we finally start treating AI Safety, AI Governance, Trust & Safety, and Fraud as interconnected disciplines rather than separate silos — something I wrote about here last year.

Velikić sees "T&S teams firmly on path to becoming risk intelligence orgs," while Mark Frumkin, Director of Customer Success at Modulate, expects T&S to "collaborate more with other teams (User Insights, Marketing, Security, CX) as tech capabilities expand to service various use cases with one solution."

How to prepare

• Start building technical skills immediately: master Claude Cowork, gen up on Python and SQL basics, go deep on how LLMs work, download the latest data analysis tools and get hands-on experience with AI investigation platforms. You don't need to become an engineer, but you need to speak the language of managing AI systems.
• Invest in your public profile. Aim to spend 10% of your working time writing, speaking, and sharing insights, and network network network. In particular, attend events outside your immediate discipline: that could mean, fraud conferences, AI safety meet-ups and security gatherings. Learn the language of adjacent fields. 
• Don't forget to invest in your job or employer, though. Volunteer for cross-functional projects and build bridges across disciplines at your workplace. Learn to translate T&S work into revenue protection, user retention, and regulatory risk mitigation. 

Expanding regulatory enforcement

In 2025, we saw the beginning of online safety regulatory enforcement against platforms. We’re already starting off 2026 with more of that, with both Australia and the UK investigating X/Twitter (EiM #317). 2026 might be the first year we see real teeth behind the regulation.

Hugh Handeyside, Principal at Full Fathom Advisory, has a compelling prediction: "The enforcement curve for online safety regulation will be steeper than the enforcement curve was for privacy regulation at a comparable stage in its evolution." Buckle up, I guess!

Asad Ali, cofounder of Illuminate Tech and former principal technologist at Ofcom, said “the question will change from "is the safety measure in place?" to "how effectively has the measure been implemented?". Similarly for regulatory compliance — moving beyond "is there an approach?" to "how good is your approach?"”

How to prepare

• Document not just what safety measures you have, but how effectively they work. Create a matrix of metrics (and work to improve outcomes) on false positives, appeals, and time-to-resolution. 
• If you operate globally, map your compliance strategy jurisdiction by jurisdiction. The US-EU split means one-size-fits-all approaches are dead. Different regions have different expectations, and cultures, and this is increasingly becoming a political fight. 

In conclusion

I was raised with the Buddhist idea of impermanence, so allow me a brief philosophical detour.

One of its core concepts is that of emptiness, or impermanence — essentially, nothing stays the same, and that there is no solid ground beneath you. In one sense, that's terrifying but when you embrace it, you realise we’re all connected and interdependent, and changing all the time. “Emptiness” becomes spaciousness, opportunity, expansion, and connection. 

In short, change is scary when it comes to your career and livelihood. It’s doubly scary when you know that your work is important in keeping others safe, and you care about doing that work well. I feel that anxiety myself all the time.

Instead of letting myself get overwhelmed, I’m trying to think about 2026 as a year of expansion, rather than contraction. My advice is to prepare for multiple futures simultaneously: think about how you can expand your role, your network, your impact, your scope, and your skills. Embrace curiosity, learn aggressively and grow as everything changes around you.

Good luck and, as ever, let me know how it goes.

Also worth reading

Social Media are helping cults to recruit and control members (The Economist)
Why? "The internet has replaced door-to-door evangelists with online influencers" – and it's easier than ever to spread information about high-control groups.

AI companies will fail. We can salvage something from the wreckage (The Guardian)
Why? Cory Doctorow has a balanced take on AI in this article, arguing that two things can be true at once: AI growth is a bubble that will burst and there are practical applications of AI that are helpful and will stick around.

State Department threatens UK over Grok investigation because only the US is allowed to ban foreign apps (Techdirt)
Why? Honestly this whole thing with Grok is just incredibly depressing to me. It should not be this hard or this controversial to just not allow NCII or CSAM!

Why newsrooms are taking comments seriously again (New_Public)
Why? EiM's founder, Ben Whitelaw, outlines the importance of new comments sections and the communities that sprout in them.