Something I have been building quietly went public last month. I am the Executive Editor of Future in the Humanities (FITH), an open exchange platform I am constructing alongside Founding Chair Professor Iginio Gagliardone, out of the SA–UK SARChI Chair in Digital Humanities at Wits University.

We are building in public, with the kind of candour that makes institutional communications teams nervous. Our opening salvo, an op-ed by Dr Rejoice Malisa-van der Walt in Business Day on what the drone strikes on Dubai should mean for African policymakers silent on AI-powered military systems, is already resonating in South Africa’s defence research community.

What follows is an invitation behind the curtain. Gagliardone and I recently discussed Mastodon, the decentralised social media platform, and why it matters for how FITH shows up in the world. We didn’t agree on everything. That is the point.

The case for leaving

Gagliardone’s argument begins with sovereignty. When an African academic, politician or activist posts on X, the value of that interaction is extracted and monetised by a company headquartered in the United States. In a recent Forbes Africa feature by Tamsin Mackay on Africa’s AI divide, he pointed out that the United States Clarifying Lawful Overseas Use of Data Act, aka the US CLOUD Act, compels American-headquartered companies to share stored data on demand from US law enforcement, even when that data sits outside the United States. Data sovereignty is not an abstraction. It is a live policy concern.

Mastodon offers a structural alternative. It is not a platform in the conventional sense but a protocol: a federated network of independently hosted servers communicating through ActivityPub. If a university, government department or any African institution hosts a Mastodon instance, conversations remain on that server. No extraction. No algorithmic curation designed to maximise engagement at the expense of wellbeing. No billionaire owner whose political allegiances might reshape the terms of participation overnight.

Robert W. Gehl, in his 2025 book Move Slowly and Build Bridges (Oxford University Press), frames the Fediverse (the broader decentralised network Mastodon belongs to) as a covenantal project: small, self-governing communities connecting through shared ethical commitments rather than commercial imperatives.

Gagliardone takes that further. Mastodon, he contends, is “by definition not a hegemonic project.” It does not aspire to replace X. It aspires to demonstrate that a plurality of ways of being online can coexist. For him, as for Gehl, this is not a platform preference. It is a political position.

The case for staying

I find the argument compelling and insufficient in almost equal measure.

I have gone on record, including on camera in TechTides: The Take, as being an active user and, yes, a fan of LinkedIn and pre-acquisition Twitter, with serious qualifications. LinkedIn is the only platform where I’ve built professional relationships that translated consistently into real opportunities. Not followers. Not clout. Actual connectivity.

African Tech Roundup led to BBC engagements, international consulting and a career-defining introduction for my wife Sithabiso that emerged from a conversation initially unrelated to her work. Pre-X Twitter gave me verified credentialisation back when it wasn’t purchasable for USD 8 a month.

These outcomes happened (at least in part) because the platforms had scale, reach and, in LinkedIn’s case, professional norms that made predictably unpredictable connections possible. Mastodon has fewer than 700,000 monthly active users globally. It peaked at roughly 2.5 million in late 2022 and has been declining since.

A platform that purposefully doesn’t keep you longer than you want to stay is admirable. But a platform nobody joins is effectively an empty room with excellent house rules.

FITH is showing up on LinkedIn, with more spots on the web set to go live in the coming weeks (think Substack and podcast vibes). We are, however, decidedly not on X — nor do we intend to be. That is the founding Chair’s line in the sand, rooted in the sovereignty and extraction arguments above, and I take it seriously. But the tradeoffs deserve naming.

LinkedIn’s recent algorithmic overhaul, well analysed by brand strategist Justin Oberman, now filters content through AI-driven semantic embeddings before your network sees it. Original thinking that does not match established engagement patterns becomes structurally invisible.

We’re poised to build a great deal of FITH’s public presence on a platform that may be quietly working against the kind of work we produce. Consistently reaching followers who sign up to our page, and new audiences who don’t yet know they’d benefit from our feed, gets harder when an algorithm decides what exists before a human being ever sees it. Unless, of course, we cough up cash to LinkedIn for the privilege of targeting exactly who gets to view our posts.

The governance question

My deeper reservation about Mastodon concerns what happens once you’re inside. Gagliardone describes its community governance as the equivalent of basic civility: you do not need to be told not to spit on someone’s floor when you walk into their home.

I am less sanguine. At what point does healthy community moderation tip into doctrinal enforcement? Gehl’s book documents this directly: the politics of defederation, instances blocking other instances over ideological disagreements, the burnout that accompanies collectively governing a space where the rules are written by the inhabitants. These aren’t hypothetical risks. They’re documented features of the system.

Gagliardone argues that absolute freedom online is overrated and points to X as evidence. I partly agree. Freedom is, after all, an unwieldy and inconvenient currency. The truth is, though, that the brokenness visible on X is not solely a product of its design, or even a product of its notorious overlord Elon Musk. It reflects something about human nature that no protocol can engineer away.

The question is whether, in correcting for the worst of platform capitalism, we create alternatives that demand a level of ideological buy-in the average user has neither the time nor the inclination to provide. Or worse, we unwittingly trade one worryingly fickle overlord for another.

Where this leaves us

Mastodon’s argument appears strongest when directed at institutions. I reckon a university or government department with the resources and mandate to host an instance and steward a community has a genuine case to make.

Gagliardone has plans to experiment with exactly that: testing the viability of Mastodon deployment in an institutional-led setting through collaborations orchestrated by the Chair. You best believe I will be riding shotgun to bring you updates on how that goes.

For individuals building careers and livelihoods across the African continent, the calculation remains different. The least convenient option often ends up being the most constructive one. But it has to be accessible before it can be constructive.

In the meantime, Future in the Humanities (FITH) will continue to build in public, on the platforms where our audience already is, while interrogating the terms on which those platforms operate. This isn’t a manifesto. It’s a working position, subject to revision.

Disclosure: Masuku also serves as strategic advisor to Professor Gagliardone’s SA–UK SARChI Chair in Digital Humanities at Wits University.

Editorial Note: A version of this opinion editorial was first published by Business Report on 17 March 2026.

Leave a comment

In December 2023, the Nigerian military launched an airstrike on Tudun Biri village in Kaduna State. Officials called it a targeted operation against insurgents. The reality, reconstructed from satellite imagery and survivor testimonies: 85 civilians dead, primarily women and children gathered for a religious festival.

No warning was issued. No explanation followed. No one was held accountable.

Tudun Biri was a system working exactly as it has been allowed to persist: unguided, unsupervised, and ungoverned.

War, upgraded

Nigeria has been in armed conflict for over fifteen years. Boko Haram, then the Islamic State West Africa Province (ISWAP), have displaced millions and exhausted one of the continent’s most formidable militaries. In response, the Armed Forces have turned, as militaries everywhere are turning, to technology.

Unmanned aerial vehicles (UAVs), AI-enabled surveillance, and automated targeting tools are no longer speculative. They are actively deployed in Nigerian skies, over Nigerian villages. In a theatre as geographically vast as this, the case for these tools is real: faster logistics, sharper intelligence, reduced risk to soldiers.

The technology works. What surrounds it does not.

Academic inquiry

My ongoing research, conducted with colleagues across three African conflict states, examines the governance architecture that is supposed to regulate how these systems are acquired, tested, deployed, and reviewed. What we found in Nigeria was, in most meaningful respects, the absence of a governance framework entirely.

Nigeria’s Public Procurement Act of 2007 contains exemptions for “special goods,” a provision that has been systematically applied to high-value defence acquisitions, including the drones and autonomous systems now deployed in conflict zones. Defence contracts, technical specifications, and rules of engagement remain classified. Parliamentary committees struggle to scrutinise them. Civil society cannot audit them. Independent investigators must piece together what happened from satellite imagery and survivor accounts after the fact.

This is a deliberate structural choice — one that transfers decision-making authority over life-and-death systems from accountable institutions to opaque procurement processes.

When AI targeting systems ingest flawed intelligence or corrupted data, there is no human-in-the-loop mechanism to prevent tragedy. In private, military planners concede that current drone operations lack the rigorous target verification protocols standard in NATO-aligned forces.

I define this condition as an “accountability desert” — a state where high-consequence AI systems operate without the legal, ethical, or institutional infrastructure necessary for accountability.

State silence

The state’s response to Tudun Biri was telling. It rebuilt homes. It promised a hospital. It established skills centres. But it did not deliver accountability. No individual commander was held responsible for what officials quietly termed “command confusion.” No independent audit of the targeting failure was conducted.

This absence has fostered a dangerous resignation among affected communities. Technological errors are increasingly accepted as inevitable: acts of God rather than failures of governance. The state has been allowed to substitute charity for justice.

As Hamza Suleiman, an award-winning Nigerian conflict reporter who has covered the insurgency for over a decade, observes: effective drone operations require approximately USD 1 million in specialised training per air officer, for flight skills, intelligence analysis, mapping, geolocation, and target identification.

The Nigerian military’s current drone units lack anything approaching this level of investment. The result is a dangerous gap between operational hardware and human expertise, one that insurgents, often trained by foreign fighters, are quick to exploit.

Some of the autonomous weapons found on African soil — including Iranian Mohajer-6 combat UAVs — are sourced directly from Iran’s supply chain. The same class of systems currently threatening the Gulf’s most sophisticated defence architectures is already present on the continent. In some cases, non-state actors now field more capable systems than the governments attempting to contain them.

Reform

Closing the accountability desert requires more than localised reform. National policy must be brought into line with continental governance frameworks.

Nigeria signed the 2024 Responsible AI in Military Domain (REAIM) Blueprint for Action, alongside 59 other nations, committing to meaningful human control and transparent procurement. South Africa participates in the Global Commission on Responsible AI in the Military Domain (GC REAIM), a Netherlands-led initiative promoting the responsible and lawful deployment of such systems. The commitments exist on paper. The gap between those commitments and what is happening on the ground, in Nigerian skies, over Nigerian villages, is in fact a void.

Closing it requires specific, concrete action: direct access for elected officials to review classified AI defence programmes; independent third-party auditing of AI decision-making processes before and after deployment; and the codification of AI-specific rules of engagement within military doctrine that mandate human override, ensuring that personnel retain the authority to intervene when automated systems fail.

South Africa has the diplomatic standing within the African Union to champion continent-wide governance standards and prevent Africa’s co-option as a testing ground for unregulated foreign military AI. The institutional capacity for these reforms exists. What is missing is the political will to apply them to the most powerful and consequential domain of all.

The families of Tudun Biri accepted the new houses. But they, and millions of other Nigerians living under the watch of systems that cannot distinguish a religious festival from an insurgent gathering, deserve more than reconstruction.

They deserve the assurance that when a drone is deployed, it is guided by a system constrained by law, overseen by independent auditors, and operated by professionals who know that a mistake will cost more than a budget line for rebuilding.

They deserve a security architecture where the law, and indeed basic human decency, has the final word.

A shorter version of this argument was published in Business Day (South Africa) on 11 March 2026, examining what the Iran–Gulf drone strikes reveal about autonomous weapons governance on the African continent. This is the longer treatment.