Why Code Is Compiled Politics, Not Neutral Engineering

Code is not a pipe. It is a politics that learned to look like engineering.

For a brief season in the 1990s, early cyberfeminists could still speak of the internet as a place where the body might be left at the door. Avatars would dissolve gender. Networks would flatten hierarchy. That wager was never foolish; it was incomplete. The platforms that followed were built by commercial enterprises, trained on the unfiltered vernacular of the street and the barracks, and monetised by engineering the capture of human attention. Technofeminism’s correction is now unavoidable: technology is not a neutral tool waiting to be used well or badly. It is a mirror, and subsequently a megaphone, of the society that finances, staffs, and scrapes data to construct it.

When the corporate laboratories shipping foundation models remain overwhelmingly male and geographically clustered within a handful of Global North jurisdictions, the resulting systems do not merely “reflect bias” as an incidental residue. They operationalise an ideological standpoint. Training corpora remain saturated with the continuum of sexualised contempt that feminist critique conceptualised long before graphics processing units became commercially cheap. Evaluation suites still treat English-language toxicity as the universal baseline for human harm. Safety policies collapse non-consensual sexual deepfakes into a generic administrative taxonomy labelled "adult content." None of this represents an accidental glitch in an otherwise neutral apparatus; it is precisely what an entrenched political order looks like when it is compiled into binary execution.

The systemic consequence is no longer speculative. Up to ninety-five per cent of deepfake material circulating across the web comprises non-consensual sexual imagery, and ninety-nine per cent of the individuals targeted are women. What historically required a darkroom, specialized tools, and a stolen physical print now demands only a consumer application and ninety seconds of automated processing. The model did not manufacture the predatory impulse; it industrialised it, scaling distribution as an integrated feature available to anyone with bandwidth. The rhetoric of technological neutrality remains the institutional alibi that allows the vendor to deflect culpability toward the end-user while the underlying system architecture generates recurring yields.

Abandoning the myth of neutrality is not a reactionary call to dismantle computational tools. It is an imperative to govern system design parameters as binding public decisions: whose data is harvested; which specific structural harms are articulated in the model's constitution; which linguistic corpora are deployed to evaluate refusal thresholds; who populates the adversarial red-teaming units; and whose lived testimony is believed when an automated classifier overlooks a threat because it was phrased in Bengali or framed through honorific societal codes rather than overt vulgarity. These interventions are not decorative diversity gestures; they represent the raw politics of the technology stack.

If algorithmic violence against women is to be structurally dismantled rather than technologically scaled, the primary intellectual concession required of software engineers, venture capitalists, and regulatory ministries is unequivocal: code is political. Strategic dialogues, such as the one convening in Dhaka, exist precisely because this principle must still be contested within institutions that remain determined to discuss innovation in sterile isolation. Until this reality is institutionalized, every peripheral reform—cryptographic watermarking, administrative takedowns, forensic investigative training, and national action plans on Women, Peace, and Security (WPS)—will remain an exercise in mopping an inundated floor while the main valve remains wide open.

Author || Chief Innovation Officer and Director General, International Trade, Investment & Technology, Ministry of Foreign Affairs, Government of the People’s Republic of Bangladesh. He is a diplomat and innovation strategist operating at the intersection of international trade, technology investments, and AI governance, actively engaged across Digital@UNGA, education and AI, sovereign compute, and economic diplomacy.

Editor's Note || The arguments and systemic critiques articulated in this article reflect the author's own policy analysis. Published with editorial standardization to further critical discourse surrounding technology governance, algorithmic ethics, and structural gender equity.