Russia-Africa AI declaration puts Lesotho’s youth at the heart of a sovereignty fight But the work starts at home

MAMELLO MARY-ALMA MOLEFI
MASERU — When nearly fifty African heads of state and
government gathered in the Grand Kremlin Palace last week for
the Third Russia-Africa Summit, the headline outcome was not
a trade figure or a security pact.
It was a single sentence buried in a new declaration – civilian
artificial intelligence is a matter of national sovereignty.
For Lesotho – a small, landlocked country with a youth
unemployment rate that official budget analysis puts at 38.9
percent – that sentence is either a historic opening or another
filing-cabinet promise.
The difference will depend entirely on what government does
next.
The summit adopted three documents: the Declaration of the
Third Russia-Africa Summit, a Declaration of Cooperation in
the Civilian Application of Artificial Intelligence Technologies,
and an Action Plan for 2027-2029. The AI declaration is the first
continent-wide instrument of its kind.
It commits signatories to equal, non-discriminatory access to the
resources needed for AI development – knowledge, datasets,
computing power, software and hardware – and to the principle

that AI developers must comply with the national laws of the
countries where their technologies are used.
It calls for reducing the digital divide through joint programmes,
knowledge-sharing and capacity-building to strengthen
“sovereign AI capabilities and technological independence.”
And it explicitly links AI development to decent work, quality
jobs, equal opportunities, training, reskilling and digital literacy.
For Lesotho’s youth, those clauses are not abstract. They
describe the difference between exporting raw data to foreign
cloud providers and building a domestic skills base that can
process Sesotho-language information, host it locally, and sell
AI-enabled services back to the region.
What Moscow actually pledged – And what it left open
The Russian framing was unusually blunt. Andrey Akulchev,
acting director of the Department of International Information
Security at Russia’s Foreign Ministry, warned that unregulated
AI could turn the digital divide into “fertile ground for modern
forms of colonialism.”
“Africa must not become either a raw-material appendage of the
new AI-based economy or a free data warehouse for those who
train their models on it,” he said.
The declaration, he added, enshrines each country’s right to
determine its own AI policy and supports multilingual solutions
including Swahili, Yoruba, Hausa, Amharic and Zulu.
That list matters. Sesotho is not on it. The omission is a
reminder that continental declarations, however sympathetic, do

not automatically include every African language or every small
state’s data needs.
Sovereignty, in practice, means building the capacity to label,
structure and host Sesotho data oneself – not waiting for a
multilingual model to arrive from Moscow, Beijing or Silicon
Valley.
Russia also pledged concrete resources. First Deputy Minister of
Economic Development, Maxim Kolesnikov, said Moscow has
been supporting 13 AI research centres employing more than 2
000 scientists since 2021 and invited African colleagues to
cooperate on the basis of those laboratories.
The declaration itself welcomes joint programmes and
knowledge-sharing. But research-centre cooperation is not the
same as a funded national data centre in Maseru. The pledge is
real; the gap between a pledge and a payroll is where Lesotho’s
strategy must live.
Lesotho’s AI foundation is real – But thin
Lesotho is not starting from zero. The government has an
Artificial Intelligence Policy of 2025, led by the Ministry of
Information, Communications, Science, Technology and
Innovation, which sets a national framework for responsible AI
development and prioritises governance, capacity-building,
digital infrastructure, sectoral integration and ethical safeguards.
In July, Deputy Prime Minister Nthomeng Majara told the UN
Global Dialogue on AI Governance in Geneva that the policy is
aligned with the Africa Continental AI Strategy and that for
small states “the central question is how countries build the

capacity, data systems, institutions and infrastructure required to
participate meaningfully.”
Lesotho proposed a Framework for AI-Ready Unstructured and
Fragmented Data, arguing that many African countries “do not
lack data; they lack data ready for AI.”
Lesotho has also joined the World Artificial Intelligence
Cooperation Organization (WAICO) as one of 29 founding
members, signing the agreement in Shanghai in July. That
membership gives Maseru a seat at a China-led governance table
and potential access to training and cooperation programmes.
But neither the Russia-Africa declaration nor WAICO
membership substitutes for the unglamorous work of digitising
clinics, standardising school records, and training young people
to build and maintain local models.
The youth employment case is urgent
The numbers are stark. The 2025/2026 national budget analysis
puts youth unemployment at 38.9 percent – 40.8 percent for
young women and 37.1 percent for young men – with 371 944
young people unemployed. The formal sector employs fewer
than 88 000 people; the informal sector absorbs more than 560
000.
A declaration that links AI to “quality jobs” and “reskilling” is,
therefore, not a nice-to-have. It is a labour-market intervention
waiting to be implemented.
The current default is data export. When a school in Quthing
sends handwritten Sesotho exam papers to a foreign API for
marking, or a clinic in Mokhotlong types patient notes into a

foreign cloud service, Lesotho is exporting the raw material of
the AI economy – and paying for the privilege in foreign
currency.
The Russia-Africa declaration’s warning against becoming a
“free data warehouse” describes this dynamic precisely. The
alternative is to build the skills and infrastructure to process that
data locally.
What Lesotho must do now
First, operationalise the declaration. Maseru should publish a
short, time-bound national action plan that translates the Russia-
Africa AI declaration into specific commitments: a local data-
hosting pilot, a Sesotho-language data-labelling programme, and
a target for the share of public-sector AI processing that stays
within Lesotho’s jurisdiction.
The declaration’s emphasis on national laws and data protection
gives Lesotho a legitimate framework to demand that foreign AI
providers comply with local rules. That framework should be
used, not shelved.
Second, invest in the missing middle—skills. The free Certificate
in Artificial Intelligence for Africa offered by Carnegie Mellon
University Africa (CMU-Africa) is the most immediately scalable
tool available.
It is a six-course online programme – covering AI foundations,
AI and machine learning in Africa, Python for AI, mathematical
foundations, machine learning essentials, and applied generative
AI—cost-free for qualified applicants, requiring only a laptop
and internet.

The programme is designed for learners who want to build AI
applications relevant to the African context, and it awards digital
badges per course and a verifiable certificate after all six. For a
young Mosotho with linear algebra, probability and basic
programming, this is a zero-cost pathway into AI work without
a visa or a flight to Moscow or Shanghai.
The government, LNDC and the National University of
Lesotho should be actively recruiting candidates and co-funding
the modest data and device costs that could exclude the poorest
applicants.
Third, build the Sesotho data layer. The UN statement’s
proposed framework for unstructured and fragmented data
should become a funded programme, not a talking point.
Schools, clinics and local government offices hold enormous
quantities of handwritten and paper-based information that no
foreign model can read.
A national labelling and digitisation corps – trained through
CMU-Africa-style short courses and employed on short-term
contracts – could convert that material into AI-ready datasets
hosted in Lesotho. This is labour-intensive work, which is
precisely why it can absorb young people quickly.
It also creates the raw material for local models that serve
Sesotho and other Southern African languages, a regional
market of tens of millions of speakers.
Fourth, use WAICO and the Russia-Africa framework as
bargaining tools, not substitutes. Lesotho’s WAICO founding
membership gives it a channel to request training opportunities
and technical cooperation from China.

The Russia-Africa declaration gives it a channel to request
research collaboration and access to Russian AI resources.
Government should table specific requests: a Sesotho language-
model pilot, a local computing facility, and a train-the-trainer
programme for Lesotho’s TVET institutions. Small states get
more from multilateral memberships when they arrive with
concrete project proposals rather than general statements of
support.
Fifth, measure what matters. Lesotho should track three
indicators annually: the share of public-sector AI processing
hosted domestically; the number of young people certified in AI
and data skills; and the value of data-processing services
exported to other Sesotho- and Zulu-speaking markets. Without
measurement, sovereignty becomes a slogan.
The sovereignty test
The Moscow declaration is a genuine diplomatic achievement. It
names the problem – AI colonialism, data extraction, the digital
divide – in language that African governments have been using
for years but rarely heard from a major power. It links AI to
jobs and skills.
It asserts the right of every country to set its own AI policy.
Those are useful foundations.
But declarations do not employ young people. Research centres
in Russia do not digitise Lesotho’s clinic registers. A WAICO
seat in Shanghai does not label a Sesotho exam paper.
The work that will determine whether Lesotho’s youth benefit
from the AI economy is the work done on laptops in Maseru,

Quthing and Botha-Bothe – building local datasets, training local
models, and selling local services.
The Russia-Africa summit has provided a sovereignty
framework. The CMU-Africa certificate has provided a free
skills pathway. The remaining variable is political will in Maseru.
If Lesotho treats the Moscow declaration as an invitation to
build rather than an excuse to wait, its young people have a
genuine route into the AI economy.
If it treats it as an achievement to be filed, the data warehouse
will remain open for business.