Civil society · Field-building · Democratic Republic of Congo

The DRC powers frontier AI with its minerals. Its people should shape what comes next.

AI Safety DRC is a civil-society initiative building AI safety and governance capacity in the Democratic Republic of Congo: the people, the networks, and the evidence that let the country engage frontier AI on its own terms.

30 September 2026: Anthropic's supported-countries list now includes the DRC. See the record.
~72%
of the world's cobalt is mined in the DRC
~42%
of global tantalum comes from the DRC
100M+
people live in the Democratic Republic of Congo
30 Sep 2026
the DRC joins Anthropic's supported-countries list

Why this work exists

The Access Paradox

The minerals inside the AI age - cobalt and tantalum in batteries, devices and data-center hardware - come in large part from the DRC. Yet until September 2026, the DRC was the only African UN member state absent from Anthropic's supported-countries list: the country feeding the compute, locked out of the models. The December 2025 U.S.-DRC Strategic Partnership on critical minerals made the imbalance a matter of policy. Access is now opening. The harder work begins: making access safe, useful and real on the ground.

A channel problem, not a demand ceiling

The DRC has leapfrogged before. It skipped fixed-line banking and went straight to mobile money; it skipped grid power and went to solar-hybrid. Low AI usage today says nothing about appetite - it describes a missing channel. Once the channel opens, adoption follows, and so does risk. Safety capacity has to arrive with access, not after it.

Access is a safety question

In our testing, frontier models fail on safety-critical tasks in Swahili - a language much of the country works in. Performance in Lingala, Tshiluba and Kikongo is an open question no one is currently resourced to answer. Deployment into the DRC without local-language evaluation is deployment with the instruments off.

A living record

The road to access

How the DRC went from excluded to listed, in three weeks. We maintain this timeline as a public reference - it is the record we wished existed.

3 September 2026

#ClaudeForDRC

Congolese digital-rights nonprofit CENSWEB launches the #ClaudeForDRC campaign, asking Anthropic for a clear path to responsible Claude and API access for the DRC.

22 September 2026

The ministry announces access

The DRC's Ministry of Digital Economy announces that Claude and the Anthropic API are being made available in the country. Reported by We Are Tech Africa.

30 September 2026

Anthropic's official list updated

Anthropic's supported-countries page now lists "Congo, Democratic Republic of (DRC)" in both the Claude.ai and API sections. First verified 00:14 SAST, 30 September 2026 - eight days after the ministry's announcement.

This tracker is maintained by AI Safety DRC. Last verified: 30 September 2026. The ministry's 22 September announcement preceded the visible list update by eight days.

Our first project

The research

"The DRC Frontier AI Access Gap: Exclusion, Leapfrogging, and the Widening AI Divide"

A documented, granular case of the DRC inside UNCTAD's named "AI Divide" framework - what exclusion looked like, why it happened, and what opening the channel must now deliver.

THE CASE

The access gap, documented

The verified country-list finding, a hypothesis audit of the usual explanations (FATF, OFAC, payment rails - none sufficient alone), and the minerals irony as the hook.

SAFETY

Access tied to safety

The Swahili tested-and-fails finding and the open Lingala, Tshiluba and Kikongo question - tying access directly to safety, not just commerce.

EVIDENCE

Comparators and primary voices

The DRC measured against Congo-Brazzaville, Cameroon and Gabon, and a qualitative study of the Congolese practitioner population that originated this project.

STAKES

The optionality argument

Access to frontier models as an input to the next mobile-money-style leap - making the stakes AI-specific rather than a generic development narrative.

Seeking an academic co-author

We are looking for a research partner who brings methodological infrastructure to the study. AI Safety DRC owns the access, the practitioner network, and the framing. If that division of labor fits your work, we want to talk.

Get in touch

What we do

Field-building first

AI Safety DRC works across four areas, anchored in civil-society capacity-building. Government advisory is the credible downstream outcome of that work, not the starting point.

01 - Primary

Field-building

Growing the people, networks and institutions that can do AI safety work in and for the DRC.

02

Policy & strategy

Policy analysis grounded in the Congolese reality - evidence the government and partners can build on.

03

Forecasting

Tracking what frontier AI means for the DRC before it arrives, in the languages the country speaks.

04

Advocacy

Making the DRC's case to the labs, funders and institutions whose country lists decide who gets a seat.

The advisory relationship with government is earned through the work: capacity first, counsel after.

Who we are

Founder

SK

Shabs Kalambaie

Founder, AI Safety DRC

Builder and systems thinker. His work spans blockchain-based mineral supply-chain traceability across the DRC and the wider SADC region, fintech, and enterprise AI fluency training - the intersection where the minerals economy meets the models it powers.

Contact: tshibangu@aisafetydrc.com