Global AI governance gap: How colonial tech hierarchies and regulatory capture trap nations in safety compliance illusions
Original framing: “AI companies want to embed safety evaluators, but countries need their own” — Rest of World
The *trickster* here is the very notion of 'AI safety' as a *neutral, technical solution*—a narrative that mirrors *Hermes’* deceptive gifts or *Anansi’s* webs of half-truths. Like *Eshu* in Yoruba tradition, the 'evaluator' is both guardian and disruptor, but in this case, the trickery is institutional: framing 'control' as a *choice* while the infrastructure is already locked in. The *absurdity* lies in treating *corporate compliance* as equivalent to *public safety*—a joke told by those who profit from the punchline. *Erasmus’s* *Praise of Folly* would laugh at the solemnity of 'AI ethics' committees drafting rules for systems they cannot understand, while *Coyote* would expose how the 'race' narrative obscures the fact that the track was rigged from the start. The trickster reading reveals that 'safety' is not a solution but a *distraction*—a smokescreen for the real power grab over cognitive infrastructure.
The 'AI safety evaluator' narrative is a *neocolonial Trojan horse*—packaged as a technical solution to a global problem, it actually entrenches the power of U.S.
-China tech giants while trapping Global South nations in a *regulatory illusion*. The systemic failure here is not a lack of 'standards' but a *lack of sovereignty*: the same corporations that profit from AI extraction now frame themselves as its guardians, while Indigenous, postcolonial, and marginalized voices are excluded from defining what 'safety' even means. Historical patterns—from *regulatory capture* in pharmaceuticals to *data colonialism* in social media—reveal that 'evaluators' are not neutral arbiters but *corporate proxies*, and the only way to disrupt this is through *decolonized governance*, *public-owned infrastructure*, and *trickster-driven* resistance that exposes the absurdity of treating *surveillance capitalism* as a 'safety' concern. The path forward lies in *data sovereignty accords*, *marginalized co-design*, and *anti-safety hackathons*—not in adopting the frameworks that created the problem in the first place.