The Sovereignty Clause

Ethiopia has signed into a China‑convened order for governing artificial intelligence. Its founding charter promises “respect for sovereignty.” The signature raises a harder question accountability to whom?

BY E. FRASHIE

Belete Molla Getahun among the WAICO signatories in Shanghai

On the sixteenth of July, on the eve of the World Artificial Intelligence Conference, a representative of the Federal Democratic Republic of Ethiopia added a signature to a document in Shanghai. With it, the country became a founding member of the World Artificial Intelligence Cooperation Organization — WAICO — an independent intergovernmental body that will be headquartered in that city and that proposes, in effect, to help write the rules by which the coming century’s most consequential technology is governed.

The domestic coverage that followed reached, predictably, for the language of arrival: a nation taking its place, at last, in the race for the future. It is a flattering frame, and it is not entirely wrong. But it obscures more than it reveals. What Ethiopia signed in Shanghai is not a lottery ticket in a technology race. It is a diplomatic alignment and, read closely, a set of propositions about power, sovereignty and accountability that deserve scrutiny rather than applause.

The Company of Signatories

Begin with the room. Twenty‑nine states signed the founding agreement; China’s foreign minister, Wang Yi, signed on behalf of Beijing, and the United Nations Secretary‑General attended the ceremony. The founding membership, by the organisation’s own account, includes Russia, Belarus, Serbia, Cuba, Brazil and Venezuela alongside ten African and twelve Asian states. What is conspicuous is not who signed but who did not. No major Western democracy not the United States, the United Kingdom, Germany, France, Japan or Australia put its name to the document.

A coalition is defined as much by its absences as by its members. This one spans four continents and, by some measures, a combined population that dwarfs the Group of Seven. Its members are bound less by a shared creed than by a shared grievance: a conviction that the rules for artificial intelligence should not be dictated from Washington or Brussels. That grievance is real, and this bears saying plainly not unfounded.

The Architecture of an Order

Here the political science becomes interesting. AI governance is not a blank field; it is what scholars call a regime complex a crowded landscape of overlapping bodies, declarations and standards. Into that landscape WAICO introduces something genuinely novel. A recent scholarly mapping of fifteen governance instruments concluded that WAICO’s design unites three features no existing multilateral body currently combines: membership open to any sovereign state, no test of values or regime type for entry, and an agenda built around development and the global capability divide.

Consider what that means by contrast. The Western‑led bodies gate entry by shared values and organise themselves around rights and safety. The universal United Nations instruments are open to all but anchored in human‑rights obligations. WAICO removes the values test altogether. Belonging becomes a function of statehood, nothing more. To a government weary of being lectured, this is precisely the appeal. It is also precisely the concern.

Sovereignty, invoked in the charter, is the very thing the arrangement quietly complicates.

For the removal of a values test is not a neutral act. It reflects a particular philosophy what Beijing has long promoted under the banner of “cyber sovereignty”: the idea that the digital domain is an extension of the state, and that its governance is the sole province of governments, unencumbered by external human‑rights obligations or by accountability to independent stakeholders. Analysts warn that WAICO threatens to carry that philosophy from the internet into the domain of artificial intelligence. And institutions matter here not merely for the ideas they profess but for the machinery through which those ideas spread the convening of officials, the training programmes, the technical standards and the research networks that quietly harden into norms.

A former Indian permanent representative to the United Nations named the anxiety in a single image: an order in which the same power both writes the rulebook and umpires the game. Whether that is prophecy or polemic, it states the structural problem honestly. Who makes the rules, and who adjudicates them?

The Word Itself

Which brings us to the word written into the charter. Among WAICO’s stated principles benefiting the people, safety and controllability, fairness, open cooperation sit “respect for sovereignty” and its companion, non‑interference. For a state like Ethiopia, jealous of its independence and proud of never having been colonised, no word could be more welcome.

And yet the paradox is difficult to miss. Ethiopia’s digital sovereignty, in any material sense, is already heavily mortgaged. The country’s first 5G network was rolled out on Huawei equipment. Its national information‑security architecture was built with Huawei’s assistance. It took on one of the largest Chinese telecoms loans on the continent to lay the fibre beneath its digital ambitions. The African Union headquarters in Addis Ababa, a short drive from the ministries now celebrating this signature was the site of a data breach, reported to have run from 2012 to 2017, that remains one of the few documented cases cited as linking Chinese‑built infrastructure to intelligence collection.

So when the Prime Minister declares, as he did last year, that Africa “must not be a passive recipient of AI tools developed elsewhere,” the sentiment is admirable and the setting ironic. One does not escape dependency by joining a standards body convened in Shanghai and headquartered there, running on a backbone financed and built by Chinese firms. That may be a rational bet — the terms are cheaper, the lectures fewer but it is a choice about whose dependency to accept, not an exit from dependency itself.

The Accountability Gap

The sharpest questions, though, are domestic and they concern accountability. WAICO’s charter speaks the vocabulary of virtue: artificial intelligence that is “people‑centred,” “safe,” “controllable,” “for good.” These are fine words. But a principle is only as good as the mechanism that enforces it, and here two problems converge.

The first is that the body itself appears to have few teeth. Writing in Nature, the science journalist Elizabeth Gibney judged that WAICO is unlikely to govern the AI industry in any enforceable way. Its instruments are convening and coordination, not compulsion. Norms without enforcement are aspirations, and aspirations bind no one.

The second problem is closer to home. Consider the state that has signed this pledge to “people‑centred” artificial intelligence. It is a state that has reached, more than once, for the internet’s off‑switch during moments of unrest. It is a state whose security apparatus has built precisely the kind of surveillance capacity that a genuinely people‑centred technology policy would exist to restrain. It is a state that, as this publication has documented, has jailed journalists. The tools of artificial intelligence facial recognition, predictive policing, the automated parsing of a population’s communications are not neutral instruments awaiting good intentions. They are force multipliers for whichever hand holds them.

When an Ethiopian citizen is wronged by an algorithm, to whom, exactly, does she appeal?

Ask, then, the question the celebratory coverage does not. When a citizen is misidentified, mis‑scored or surveilled by a machine, to whom does she turn? Not to WAICO, which claims no such mechanism and no such role. Not, on present evidence, to a domestic framework robust enough to constrain the state that deploys the technology against her. The charter’s promise of “people‑centred” AI, absent any tribunal to which the people may bring a grievance, is a promise addressed to governments about how they intend to describe themselves not to citizens about what they may demand.

The Minister and the Prisoner

Every cabinet tells a story about power, and Ethiopia’s tells one worth reading closely. The portfolio that carries the nation’s digital ambitions the ministry from which its artificial‑intelligence agenda, its “five million coders” and its technology diplomacy all flow belongs to Dr Belete Molla Getahun. It was Belete who travelled to Shanghai this month and, beneath the Ethiopian tricolour, set his signature to the WAICO agreement on the republic’s behalf a moment he shared, days later, on his own Facebook page. And Belete is not a man of the ruling Prosperity Party. He is the founder and chairman of the National Movement of Amhara, and among the most recognisable opposition politicians in the country. His presence in that hall is, on paper, evidence of a broad tent: a government generous enough to seat its critics, and to let one of them sign for the nation abroad.

Read the arrangement more carefully, though, and it resembles less a pluralism than a transaction. NaMA five seats, and the self‑styled face of the opposition accepted a single cabinet post on the Prime Minister’s offer. One chair at the table, held by the chairman, in exchange for a voice inside a government the party was elected to oppose. Critics have since reached for a word harsher than “loyal opposition.” In this telling the technology ministry is not a share of power but the price of a party’s silence, paid in advance.

And here the arrangement acquires its cruelty. For while the chairman of NaMA sits in cabinet administering the state’s digital future, the party’s own spokesperson and most forceful parliamentary voice sits in a cell.

Christian Tadele Tsegaye founding member of NaMA, its political‑affairs chief since 2018, and, until his arrest, chair of a parliamentary committee charged with scrutinising government spending was taken from his Addis Ababa home in early August 2023, days after the government declared a state of emergency and opened its campaign against Fano fighters in Amhara. His arrest followed, by a matter of days, a speech on the floor of parliament criticising the Prime Minister and a widely shared post calling on him to resign. He was charged under the anti‑terrorism and firearms proclamations. In March 2024 the House of Peoples’ Representatives the same chamber in which he had questioned the Prime Minister to his face voted to lift his immunity; two members opposed, two abstained. He has been held at Awash Arba and later Kilinto, where, by the accounts of his lawyer and family, his health has collapsed: surgery followed by inadequate care, food turned away at the prison gate. Amnesty International has warned that Ethiopia’s emergency powers enable precisely this kind of arbitrary and prolonged detention. He is not alone; the NaMA co‑founder Dessalegn Chanie and others were detained under the same emergency.

The government’s position is that Tadele aided an armed insurrection. His supporters, and a widening chorus of human‑rights organisations, call the charges a punishment for dissent. The Ethiopian Tribune does not adjudicate the criminal case here. But it insists on holding two facts within a single frame because the government’s own cabinet arithmetic places them there.

In one frame: the opposition chairman who signed for Ethiopia in Shanghai, lending his credibility to the state’s language of “people‑centred,” “safe,” “controllable” AI. In the other: his own colleague, elected under the same banner, imprisoned and ailing, silenced by the very state the minister now represents abroad. The instruments of the digital state the surveillance capacity, the security apparatus this newspaper examined above are no abstraction to the Amhara opposition. They are the architecture of their own suppression.

Its price is one cabinet chair. Its cost is a colleague in Kilinto.

So the accountability question sharpens to a point. What is a technology minister’s pledge of people‑centred AI worth, if he cannot or will not secure the liberty of the man who stood beside him at his own party’s founding? And what does “opposition” mean, when its price is one cabinet chair and its cost is a colleague in Kilinto?

The Questions Worth Asking

None of this is to say Ethiopia was wrong to sign. The case for WAICO is serious and should not be caricatured. The digital divide is real; the gap between the Global South and the handful of states that own the frontier is widening into something structural. The incumbent Western‑led bodies have indeed been exclusionary, gating the conversation by values that too often function as a euphemism for geography. A body that offers infrastructure, training and a seat at the table to states long treated as recipients rather than authors is not obviously a trap. It may be the most rational instrument available to a country that cannot build the frontier alone.

But rationality is not the same as safety, and opportunity is not the same as sovereignty. The questions that ought to occupy Ethiopia’s policymakers and its press are not whether the country has “arrived,” but sharper ones. Whose norms will Ethiopian officials be trained to adopt, and at whose expense? What accountability will exist, to any body or any court, when the technology is turned inward against the citizen? And what does it mean to sign a sovereignty clause with one hand while, with the other, deepening a dependency that no clause can dissolve?

A signature in Shanghai is easy. The reckoning it postpones is not. The Tribune will go on asking who, in the end, is meant to be accountable and to whom.

E. Frashie writes on politics and technology for The Ethiopian Tribune.