Trust and safety is the fastest-maturing function in the digital economy. The EU Digital Services Act has been in full force across all in-scope platforms since 2024, the UK Online Safety Act has been operational since 2025, and Australia, Brazil, Singapore and Canada all have comparable regimes in 2026. The compliance bar is materially higher than it was three years ago, and the consequences for getting it wrong are now criminal in some jurisdictions, not just reputational.
That has reshaped the buyer profile. Five years ago, content moderation outsourcing was bought by performance teams optimising cost-per-decision. Today it is bought by general counsel, chief trust and safety officers, and chief risk officers — with cost as a secondary consideration to quality, defensibility and moderator wellbeing.
This guide is for those buyers.
The function has fragmented into six distinct workstreams that mature platforms now staff separately:
Most platforms outsource workstreams 1, 2, 3, 4 and the routine portion of 5, and retain workstream 6 and the strategic portion of 5 in-house. A vendor that asks to own workstream 6 is a vendor to walk away from.
Content moderation involves repeated exposure to traumatic material — child sexual abuse material (CSAM), terrorism content, graphic violence, self-harm, hate speech. The mental health impact on moderators is established science: meta-analyses since 2020 document elevated rates of PTSD, depression and burnout in unsupported moderator populations. Lawsuits against major platforms by former moderators have settled at meaningful values.
A responsible 2026 contract starts with the wellbeing programme, not the cost-per-decision. The non-negotiables we operate under, and that any responsible buyer should require:
A vendor that does not lead with this is a vendor that will eventually generate the headlines that destroy your brand.
For EU-facing platforms, DSA imposes a set of operational requirements that translate directly into how a moderation operation must be designed:
These are not features to bolt on. They are design constraints that shape staffing, tooling, queue design, audit logs, and the data warehouse. A serious moderation partner ships against these requirements as defaults; an outsourcing vendor that needs you to explain DSA to them is not the right vendor.
After thousands of moderation hours, five operational levers determine quality more than anything else.
Honest 2026 fully-loaded rates for nearshore trust and safety operations, Casablanca-delivered:
These rates are 40-55% below equivalent EU onshore moderation operations. They reflect senior staff and the wellbeing infrastructure described above. A vendor quoting €3-€5/hour for "content moderation" is not operating that infrastructure; the gap is real money funding real services that you do not want to skip.
Cultural calibration matters more in moderation than almost anywhere else in support operations. A sarcastic Spanish comment is not the same speech act as a sarcastic Arabic comment; a French regional slur lands differently in Paris and in Brussels; a meme that is benign in Brazilian Portuguese can be hateful in European Portuguese.
Casablanca staffs the following languages and dialects at moderator-grade fluency: English (UK and US registers), French (France, Belgium, Quebec via remote augment), Spanish (Spain and Latin America), Arabic (Modern Standard, Maghrebi, Levantine, Gulf), Portuguese (Portugal via Madrid hub), Italian (via Madrid hub).
For other languages — German, Dutch, Polish, Turkish, Greek, Nordic languages, Japanese, Korean, Vietnamese, Tagalog, Indonesian — we either staff via remote native speakers integrated into the Casablanca operations, or partner with regional hubs in our network. We will tell you directly which model applies; we do not pretend to staff languages we cannot staff at quality.
${RELATED}
${CTA}
Yes. DSA is operational and procedural, not geographic. A vendor that has built the audit trail, the appeals flow, the Statement of Reasons pipeline and the transparency reporting data model can comply from any jurisdiction. We have shipped against DSA for in-scope clients since 2024.
For appropriately permitted clients, yes, with mandatory hash-matching partnership integration (NCMEC, IWF, Project Arachnid where applicable), a dedicated specialist team with extended wellbeing programme, and law-enforcement coordination protocols.
A separate appeals queue staffed by senior moderators who did not make the original decision, with policy-team adjudication on disagreed decisions and feedback loops into moderator coaching.
Synthetic content (deepfakes, AI-generated CSAM, AI-generated harassment) is now a meaningful share of every moderation queue. Detection requires a combination of provenance signals (C2PA where available), platform-specific classifiers, and human review. We treat synthetic content as a distinct policy category with its own enforcement matrix.
Yes. Every client engagement includes a quarterly operations review with full audit-trail access, sampled-decision review, wellbeing-programme evidence, and policy-calibration session minutes. Independent third-party audits on request.
CALL IT DEV — Software, AI and dedicated tech teams — Casablanca | Madrid | Dubai — contact@callitdev.com — +212-537-373777