This is a shift from adjudication to pre-crime analytics . As Crawford and Schultz (2019) argue, algorithmic systems "produce suspicion rather than respond to it." The user has no right to confront the algorithm, no discovery of the training data, and often no meaningful appeal. In Jasper v. Meta (N.D. Cal. 2024), the court held that Section 230 shielded Meta from liability, but noted that "the plaintiff was effectively tried and convicted by a statistical model."
Finally, legal norms must be culturally embedded. Platforms should design friction into accusatory features (e.g., requiring a verified identity for public accusations, adding a mandatory "presumption reminder" before sharing an accusation). Digital literacy curricula should teach the distinction between suspicion and conviction. presumed innocent en ligne
The principle of presumed innocent is not a natural feature of online spaces; it is a hard-won legal achievement that must be deliberately reconstructed for the digital age. Without intervention, the default architecture of networks—automated, opaque, and instantaneous—will continue to invert the presumption, punishing first and hearing later. But with targeted procedural reforms, private and public actors can restore the essential balance: no punishment without process, and every accused remains innocent until proven otherwise. This is a shift from adjudication to pre-crime analytics
This is the purest inversion of the presumption: the burden shifts to the accused to prove their innocence in real-time, before an unbounded audience, with no rules of evidence, no right to silence, and no neutral arbiter. As noted by Citron (2014), "digital vigilantism operationalizes guilt until proven innocent." The speed and scale of social networks mean that even a later exoneration rarely restores the prior status quo. Meta (N
Private online platforms (X, Meta, TikTok) moderate billions of content items daily. Their terms of service often include clauses allowing suspension or removal "at our sole discretion." In practice, automated systems flag content based on statistical risk scores. A user is not presumed innocent; rather, a post is presumed violative if it matches a pattern (e.g., certain keywords, account age, report frequency).
The principle of presumed innocent until proven guilty is a cornerstone of modern liberal legal systems. However, the migration of social, commercial, and judicial activities to online platforms (en ligne) has fundamentally destabilized this principle. This paper argues that digital environments—from social media moderation to algorithmic surveillance—systematically invert the presumption of innocence, replacing juridical due process with probabilistic risk management. By examining three distinct online spheres (private platform governance, criminal procedure involving digital evidence, and public discourse), this paper demonstrates that the classical presumption is neither technically nor culturally native to the digital space. It concludes by proposing a hybrid framework of procedural safeguards adapted to network architecture.
In analog systems, this presumption is enforced through gatekeepers: judges, rules of evidence, cross-examination, and public pronouncement of guilt only after conviction. The key insight is that procedure precedes punishment . No legitimate deprivation of liberty or reputation occurs without a prior adversarial process.