Editorial
Despite the global ceremony and ostensible triumph of India’s AI Impact Summit, the event was overshadowed by farce and humiliation that exposed the vacuity of its grandiloquent ambitions: a Galgotias University pavilion was compelled to withdraw after a commercially purchased Chinese robotic dog was mortifyingly paraded as an indigenous innovation—a debacle that went viral and cast doubt on the summit’s claims of authentic technological prowess—and, in a theatrical display of political theatre, members of the Indian Youth Congress staged a shirtless protest within the venue, brandishing slogans that echoed through the halls as foreign delegates looked on in stunned disbelief. In this climate of spectacle masquerading as policy, the Delhi Declaration on Artificial Intelligence—signed, with ritualistic unanimity, by 88 countries including Sri Lanka—was hailed as a “successful attempt” at framing a global consensus on AI governance, yet such praise rings hollow when a summit intended to project technological leadership is remembered less for its diplomatic language than for its comical missteps.
It was, in the argot of multilateral diplomacy, a performative concordat—an instrument less juridical than theatrical, less enforceable than hortatory. The so-called declaration, circulated under the auspices of the Ministry of External Affairs (India), is replete with the familiar incantations of transnational cooperation, invoking phrases such as “AI for All” and the aspiration to treat artificial intelligence as a “shared global good”. Yet such locutions, however mellifluous, possess the normative density of vapour. They neither instantiate binding obligations nor create adjudicable rights; they are the diplomatic equivalent of a gentleman’s agreement in an age no longer populated by gentlemen.
International law, as any publicist of the old school will attest, distinguishes between lex lata—the law as it exists—and lex ferenda—the law as some committee wishes it to be. This declaration belongs emphatically to the latter category. Its clauses are drafted in that curious idiom of consensual ambiguity designed to offend no sovereign ego while committing none to measurable conduct. States “encourage”, “endeavour”, and “seek to promote” frameworks that remain conveniently undefined. There is no compliance mechanism, no supervisory tribunal, no sanction for delinquency. One is reminded less of a treaty regime than of a university mission statement laminated and forgotten on a corridor wall.
The declaration’s champions proclaim that it “democratises access” to technological capability. Democratisation, however, presupposes parity of agency. In a world where computational infrastructure is cartelised by a handful of jurisdictions and corporations, the rhetoric of inclusivity risks degenerating into what critical jurists might term a simulacrum of consent. When the text extols “equitable participation of all nations in the AI ecosystem”, it elides the brute asymmetry of semiconductor supply chains, data capitalisation, and proprietary algorithms. Equity, without redistribution of technological capital, is merely etiquette.
Consider the geopolitical dramaturgy underlying the signatures. That both the United States and China could append their imprimatur to the same parchment is heralded as evidence of a nascent global consensus. A more sceptical observer might characterise it as convergence at the lowest possible denominator—an agreement so anodyne that even strategic rivals could endorse it without forfeiting advantage. Consensus, here, is not proof of shared conviction but of shared convenience.
The declaration assures us that AI must remain “human-centric, trustworthy, and responsible”. Such tripartite formulations are the stock-in-trade of what one might call ethical ornamentalism: values articulated with great solemnity and negligible operational consequence. Nowhere does the instrument delineate liability for algorithmic harm, jurisdiction over cross-border data exploitation, or restitution for automation-induced economic displacement. The absence is not accidental; specificity would entail obligation, and obligation would entail resistance.
From the vantage point of the Global South, including countries such as Sri Lanka, the declaration’s promise of technological fraternity may appear alluring. It gestures toward capacity-building, knowledge exchange, and what it terms “collective advancement of digital futures”. Yet history furnishes ample precedent for scepticism. Developmental vocabularies have long functioned as diplomatic currency through which structural inequalities are rhetorically acknowledged yet materially preserved. One recalls the proliferation of environmental declarations that celebrated sustainability even as emissions ascended with cheerful impunity.
Indeed, the very notion that AI can be “democratised” without interrogating the political economy of data verges on conceptual incoherence. Data, after all, is not an abstract commons but an extracted resource—harvested, monetised, and enclosed within proprietary architectures. To democratise AI meaningfully would require something far more radical than workshops and memoranda; it would necessitate antitrust interventions, technology transfers, and perhaps even the reclassification of certain computational infrastructures as global public utilities. The declaration contemplates none of this. Its radicalism is lexical, not structural.
The legal minimalism of the document is further underscored by its deliberate non-binding character. It is, to employ the language of the Vienna Convention, devoid of intent to create legal relations. This affords signatories maximal interpretive latitude: they may invoke the declaration when convenient and disregard it when inconvenient. Such instruments are sometimes defended as precursors to harder law, but just as often they become substitutes for it—a phenomenon scholars have described as “normative displacement”, where the appearance of action forestalls its reality.
Even the much-celebrated proposal for shared platforms and collaborative research frameworks is articulated in terms so Delphic that implementation can be postponed indefinitely. States are invited to “foster environments conducive to innovation” and to “advance interoperable governance approaches”. These are not obligations; they are aspirations drafted in the subjunctive mood of perpetual deferral. The grammar itself resists accountability.
One must therefore ask whether the declaration is less a charter for technological justice than an exercise in reputational management. In an era when democracy itself is increasingly commodified—auctioned through influence markets, data analytics, and platform monopolies—the spectacle of governments proclaiming the “democratisation of AI” carries a faintly ironic timbre. It is akin to oligarchs convening to discuss egalitarianism over a well-catered luncheon. The language of access masks the reality of control.
This is not to deny that dialogue among nations possesses intrinsic value. Multilateral conversation can temper excess, establish expectations, and generate soft norms that gradually crystallise into custom. But such evolution requires candour about power, not its polite concealment. A declaration that speaks endlessly of inclusion while sidestepping the mechanics of exclusion risks becoming what Roman lawyers might have called a nudum pactum—a bare agreement, morally suggestive yet legally void.
The venue in New Delhi thus becomes symbolic: a crossroads where ambition, anxiety, and geopolitical theatre intersected beneath the banner of technological humanism. Whether this gathering will be remembered as the embryonic stage of a genuinely plural AI order or merely another communiqué consigned to archival obscurity depends on what follows. Declarations, after all, do not change the world; enforcement, investment, and redistribution do.
Until such material commitments emerge, the New Delhi proclamation will remain an artefact of diplomatic optimism—eloquent, expansive, and faintly unreal. In a century already adept at converting ideals into branding exercises, the democratisation of artificial intelligence may yet prove to be the most sophisticated slogan of all: a promise encoded in impeccable prose, running on hardware owned by someone else.
