Segment ranking · August 2026

The IPO Trust & Safety 10

The independent outsourcers running intelligent trust & safety operations: AI in production, documented workflows, and evidence you can check. We screened 52 companies. Ten cleared the bar.

The segment, and why it matters

Trust & safety is the outsourcing segment where AI arrived first and where the stakes stay highest. Every marketplace, dating app, gaming platform, and social community needs moderation that runs around the clock, in dozens of languages, under regulatory regimes that now carry real teeth: the EU Digital Services Act, the UK Online Safety Act, and a growing list of US state laws.

The giants dominate the headlines here, and the consolidation wave is eating the specialist tier: IntouchCX bought WebPurify in January 2025, TDCX bought Open Access BPO in April 2025, and EXL announced its acquisition of iMerit in June 2026. Sama, once the most famous name in the segment, exited moderation entirely in 2023. Below the giants sits a shrinking band of independent operators who built moderation practices with production AI and, in the best cases, named clients and measured results. This list is where to find them.

The visualization

The Pivot Path

Hover or tap a dot to see the company behind it.

The Trust & Safety 10 Ones to Watch Commercial model not published Dot size = overall weighted score

The horizontal axis measures intelligence in production: AI deployment depth and the human-AI operating model, weighted as published. The vertical axis measures accountability: outcome-based commercial models and governance. The diagonal band is the path itself, where automation and accountability advance together. Every company here earned its place, so position is a prompt, not a verdict. In this segment the picture is blunt: the automation runs ahead of the accountability, and the open upper right of this chart is the leadership opportunity. The top right is the destination: Intelligent Process Outsourcing, fully realized.^*

There is harmful content online, and trust & safety teams are the frontline defense that keeps it away from children, families, and communities. That defense has always come at a human cost: the people who review the worst material absorb exposure the public never sees, a toll documented in courtrooms and clinics, including a $52 million settlement to more than 14,000 US moderators. AI in production changes the equation: every item a model resolves before a person sees it protects the reviewer. The leaders on this list already work this way: automation carrying up to 94% of moderation volume in published case studies, exposure tiered and rotated on the hardest queues, and on-site psychological support. Besedo, SupportZebra, Horatio, and ModSquad are out front on this, and their write-ups give the specifics. It is work worth honoring, and it is the standard this list holds the segment to.

The ranking

Ten that cleared the bar

Before you read the scores, here is how they came to be. We started with 52 companies and asked the same question of every one: can you prove it? Each rank comes from published evidence across the eight dimensions below. What we prized most is what a buyer would prize: AI actually running in production, named clients willing to stand behind the work, results with numbers attached, and honest treatment of the people doing the hardest job on the internet. No company paid to be here. No company can.

The top score on any dimension is reserved for evidence someone else staked their name on; silence on a fact a company controls scores as what it is; and where missing evidence depends on third parties, we rescale rather than guess. Two numeric ties survived (ranks 5 and 6, ranks 8 and 9); the published tie-break rule ordered both: third-party AI documentation first, verified certifications second.

Besedo

Weighted score 3.95

Marketplace moderation with the receipts published

Besedo is what this segment looks like when the transition is complete. The Stockholm firm has run hybrid AI and human moderation since 2002, and its Implio platform carries the load in public, dated case studies: 85% automation for marketplace Kaidee, 94% automation at 99.8% accuracy for Anibis, and moderation for Change.org's 460 million users. The client roster is named and long: Match.com, Vox Media, Zalando, Adevinta, BlaBlaCar. At roughly 400 people it outperforms moderation operations ten times its size, and it publishes the numbers that prove it. The one thing it keeps private is its commercial model, which is the only reason the score is not higher.

PartnerHero, a Crescendo company

Weighted score 3.80

The only independently documented moderation client on the list

PartnerHero is the only operator here whose moderation work has been documented by an independent investigation: the Bureau of Investigative Journalism confirmed it runs much of Grindr's content moderation. It hired a VP of Trust & Safety in 2023, acquired AI QA firm Summatti, holds ISO 27001, and now delivers on Crescendo's augmented-AI platform with outcome-based pricing. The same investigation that proved the client relationship also reported Honduran moderators averaging about $700 a month, and that fact costs it heavily on the workforce dimension. The capability is real and independently verified. The pay question deserves a direct answer, and we will ask for one.

Hugo

Weighted score 3.25

A real T&S practice with the segment's best workforce story

Hugo pairs a real trust & safety practice, with policy matrices, escalation workflows, and user verification, with the strongest workforce story in the segment: the Hugo Academy has trained more than 2,000 young people in underserved communities across its African delivery network, placements are two-thirds women, and the company reports 98% agent retention. Clients include TikTok, Hilton, and Faire. Its GenAI operations work runs adjacent to moderation delivery. What we could not find is a public outcome-based commercial model, and its moderation AI evidence is thinner than its community credentials.

LiveWorld

Weighted score 3.17

Pharma social moderation nobody else can touch

The most specialized operator on the list. LiveWorld has moderated brand communities since 1996 and owns a niche nobody else here can enter: pharmaceutical social moderation with FDA adverse-event handling built into the workflow, roughly 90% of its revenue concentrated in healthcare across 375+ digital properties in 70+ country-language combinations. Its LiveInsight AI launched in June 2026. Our verification pass found a profitable business with $7.3M in cash behind the microcap ticker, and a commercial model moving to enterprise licenses, documented and priced but tied to access rather than outcomes. The smallest company here, and the deepest in its vertical.

SupportZebra

Weighted score 3.15

Certified moderation delivery, third-party AI documentation

A Philippines-delivery operator with the segment's best third-party AI documentation: Zendesk, whose case study confirmed generative AI in production for real-time sentiment analysis and auto-summarization. Its dedicated moderation line covers text, image, and video with SOC 2 Type 2 and PCI-DSS underneath, and a 94% employee recommend rating on top. We verified the detail: the Zendesk-documented AI runs in its contact-center delivery, and its moderation line runs human-in-the-loop, integrating with clients' own AI stacks, which is a legitimate operating model in this segment and we describe it as exactly that. Its moderator-care practices are published in first person: counseling access, scheduled rotation on high-risk queues, and content exposure tiered by tenure.

Horatio

Weighted score 3.15

Flagship No. 4 brings its T&S line nearshore

Horatio brings the same strengths that earned its flagship rank: an AI implementation practice with named partners in Ada and Loris, the top workplace ranking in its sector in the Dominican Republic, and 441 internal promotions in a single year. Its moderator-care story is the segment's most concrete: free on-site access to licensed therapists and psychologists, with rotation across exposure levels, corroborated by independent press coverage of its Santo Domingo campus. Its moderator-care story is the segment's most concrete: free on-site access to licensed therapists and psychologists, and rotation across exposure levels to prevent burnout, corroborated by independent press coverage of its Santo Domingo campus. Its moderation offer spans retail, finance, healthcare, and gaming. The T&S-specific evidence runs behind its CX evidence: no named moderation clients, no published moderation metrics, and no public outcome model. The trajectory says those gaps close; the score reflects that they have not yet.

ModSquad

Weighted score 3.10

T&S-native since 2007, named logos, certified platform

ModSquad was founded as a trust & safety firm in 2007, before the term was fashionable, and its service depth shows it: moderation, policy development, fraud ops, user appeals, and AI safety work including red teaming and RLHF, delivered by contractors in 90+ countries for named clients including Spotify, EA, Vimeo, and the NFL, on a SOC 2 Type 2, HIPAA, and PCI-certified secure platform. Two things hold it at seven: it bills hourly, which is the seat model in different clothes, and its Glassdoor reviews rate compensation at 2.3. The distributed-contractor model is disclosed here and weighed in the write-up. It is also the only company here with an explicit public claim that AI review protects its people, with AI pre-classification ahead of human review and a sustainability model built on project choice, short intervals, and rotation. It is also the only company here with an explicit public claim that AI review protects its people, with AI pre-classification ahead of human review and a sustainability model built on project choice, short intervals, and rotation.

Mindy Support

Weighted score 3.05

Published multilingual video moderation at livestream speed

Mindy Support publishes what most of this segment only claims: a dated case study moderating video in more than 20 languages for a large video platform at sub-60-second livestream latency. SOC 2 Type 1, ISO 27001, and ISO 9001 sit under the delivery, and a 4.8 rating across 37 verified Clutch reviews sits on top, with named clients including Bosch and MacPaw. Founded in Ukraine, it kept operations running through the war. Under our source-quality ladder the case study scores as self-published and anonymous because the video platform is unnamed; a named client would restore the higher score.

5CA

Weighted score 3.05

The gaming player-safety specialist

5CA has run player support and community safety since 1998, remote-first across 95 countries, with named clients including CCP Games and Psyonix and a COPC-certified operation. Its AICX venture points the right direction on agentic AI. What keeps it at nine is evidence: the AI claims lack named client cases, and per-agent contracts anchor the commercial model to time rather than outcomes.

SupportYourApp

Weighted score 2.95

Moderation in 60+ languages under a serious security posture

Flagship No. 18 delivers moderation in more than 60 languages covering fraud detection, abusive content identification, and community guideline enforcement, with AI-powered tooling supplementing human judgment on complex cases and a serious security posture underneath. The moderation line is real and marketed with specifics. Named moderation clients, production metrics, and a published commercial model are what separate it from the honorees above it.

Ones to Watch

Real trust & safety delivery that missed the ranking on evidence, in score order where scored: Oworkers (2.95, productized AI pre-filter pricing published openly, a rare transparency), ContactPoint360 (2.80, T&S named but depth unproven), Foiwe (2.78, North American clients verified, governance certifications still unverified), Conectys (2.50, twenty years of UGC moderation at scale), Helpware (2.50), Enshored (2.39). Oworkers ties the No. 10 score at 2.95; SupportYourApp holds the tenth spot on the second standing tie-break, verified certifications. Unscored pending specific gaps: The Social Element, SupportNinja, DDD.

Companies in this band can carry a featured buyer profile here. That placement is paid, always labeled, and never changes a score or creates a ranking. Made the shortlist, not the list: we think that is worth saying out loud.

Featured solutions · Sponsored

Ranked where it matters to your platform

Retail & marketplaces: the ranked five

Re-weighted with vertical depth at 20%. Listing moderation, review integrity, counterfeit screening, seasonal peak.

1Besedo4.16

2Horatio3.33

3Hugo3.22

4SupportZebra3.13

5Mindy Support3.04

BFSI: Ones to Watch

We say this plainly: fewer than five independents cleared our gate with citable BFSI-specific T&S evidence, so per our published rule this is a watch band rather than a forced ranking. Watch: SupportZebra (fintech tier-1 with SOC 2 + PCI), Horatio, Oworkers (verified regtech delivery), ModSquad (fraud and risk ops). The first independent to publish a named BFSI moderation case study with measured results takes this category next year.

Healthcare: Ones to Watch

Nearly a category of one. LiveWorld is the standout: adverse-event handling built into pharma social workflows across 375+ healthcare properties. Also watch Hugo (HIPAA-trained moderation staff), Horatio, and SupportZebra.

The trend read

Three things are true at once

First, consolidation is removing independents faster than new ones appear: three specialist acquisitions in eighteen months. Buyers who want an independent partner have a shrinking list, which makes this ranking more useful and shorter-lived than most.

Second, the AI transition here is real and asymmetric: the best operator we scored automates 85 to 94 percent of volume in production while the median candidate still markets "AI-powered" with no named tool. The gap between claim and evidence is wider in trust & safety than in any segment we have covered.

Third, commercial models tied to performance are still rare. Only three of the ten honorees publish how they charge, and most of the segment still sells time. This is where a provider can demonstrate real leadership: the first mid-size independent to price moderation on outcomes, with an accuracy commitment behind it, will own that conversation with buyers.

Methodology

How we rank

This list covers independent outsourcers under 5,000 staff with trust & safety delivery in production for North American clients. We score eight dimensions with published weights: AI deployment depth 20%, outcome accountability 20%, governance and failure readiness 15%, human-AI operating model 10%, workforce health 10%, vertical depth 10%, client proof 10%, positioning clarity 5%. A 4 or 5 requires a citable source. Where public evidence was insufficient we mark ID. On dimensions a company controls disclosing (its commercial model, certifications, vertical proof, positioning), an ID scores a 2: not publishing the answer is the answer. On dimensions that depend on third parties (workforce ratings, independent client proof), we rescale rather than guess, because a blind spot of ours should never hurt a company. We adopted this rule after our own testing showed pure rescaling could reward opacity, and we say that plainly too. A company carrying three or more ID flags goes back to research rather than onto the list.

The ranking above is editorial and follows the published methodology. No company can pay for placement or position. Featured profiles and partner placements beside the list are paid, and every one of them is labeled. Current or former Pivot AI Global advisory clients are marked wherever they appear; there are none on this list. The full scoring matrix with a source for every score is available for audit.

Nominations

Think a company is missing?

The hardest part of building this list was finding the companies. If a trust & safety operator belongs here and is not, make the case. It goes straight into the next review.

More lists

The IPO 25 (flagship) All lists LPO · September SDR-as-a-Service · October APAC · November