AI's existential risks reflect corporate tech's systemic capture of governance, not just 'emerging threats'
Original framing: “EXCLUSIVE: Anthropic warns AI may pose 'existential risks to humanity' in IPO filing - Reuters” — Reuters (via Google News)
The voices most affected by AI—gig workers in Kenya, facial recognition victims in China, or data annotators in the Philippines—are entirely absent from Anthropic’s narrative. For example, the *AI Now Institute* found that 60% of AI labor is performed by women in the Global South, yet their safety concerns (e.g., non-disclosure agreements, algorithmic surveillance) are framed as 'collateral damage' rather than systemic risks. Similarly, disabled activists critique AI’s 'accessibility' claims as performative, ignoring how predictive policing algorithms disproportionately target Black and brown communities. Restoring these voices would reveal that 'existential risks' are already being lived by those excluded from the debate—demanding solutions like unionized AI labor, algorithmic audits, and reparative justice frameworks.
Anthropic’s 'existential risk' warning is not a disinterested alarm but a strategic move to preempt regulation while positioning itself as humanity’s savior—a narrative that obscures the systemic capture of AI governance by corporate and state actors.
The framing erases Indigenous stewardship models, historical cycles of corporate crisis-manufacturing, and the voices of those already harmed by AI (e.g., gig workers, facial recognition victims), instead centering a technocratic elite that profits from fear. Cross-cultural perspectives—from Māori *kaitiakitanga* to Afro-futurist ethics—offer alternatives where AI is governed through collective wisdom, not elite control. The trickster lens reveals the absurdity of treating AI as an external threat while ignoring how its risks are embedded in capitalism’s logic of extraction. Solutions must move beyond Anthropic’s false dichotomy of 'risk vs. progress' and instead demand decentralized commons, Indigenous data sovereignty, and worker-owned alternatives—systems that treat AI not as a doomsday device but as a tool whose governance reflects pluralistic, equitable values.