Michael John-Hopkins, 17 July 2026 11 mins read
In July 2025 in Bradshaw and Others v United Kingdom, the European Court of Human Rights confronted the unexpected tension between freedom of expression and free elections. Democratic societies must preserve the broad freedom of political expression on which open electoral debate depends, but that freedom may also be exploited through coordinated foreign influence operations designed to undermine free elections.
The applicants argued that the United Kingdom had failed to adequately investigate, and protect against, Russian interference in its democratic processes. The Court did not find a violation. It emphasised the difficulty of measuring the effects of disinformation and influence campaigns and allowed the United Kingdom a wide margin of appreciation, particularly because no clear European consensus has yet emerged on the specific protective measures States must adopt. The judgment was nevertheless significant. It accepted that Article 3 of Protocol No 1 of the European Convention on Human Rights (right to free elections) may impose positive obligations where hostile-State interference creates a real risk of impairing the very essence of the right to free elections.
Before the judgment, the UK Parliament Intelligence and Security Committeeâs Russia report (2020) had already described the United Kingdom as a target for Russian disinformation and political influence operations. Importantly, it distinguished disinformation from the broader playbook of malign interference, which may combine manipulated narratives with illicit funding, cyber operations, interference with electoral machinery, covert amplification and âhack-and-leakâ activity. Such operations frequently seek less to secure agreement with a single message than to intensify division, confuse attribution, discredit reliable sources and erode trust in democratic institutions.
In Bradshaw, the Court accepted that there was evidence of interference in the United Kingdomâs democratic processes of sufficient intensity to be capable of impairing the very essence of the right to benefit from free elections. Its conclusion was not that the threat was unreal, but that the United Kingdomâs response remained within its wide margin of appreciation.
That tension captures one of the central problems now facing democratic legal systems. Electoral rights can no longer solely be guaranteed at the ballot box but must also be protected in the communicative environment in which electoral preferences are formed. That environment is increasingly structured by platforms, recommender systems, political advertising, synthetic media, data profiling, covert funding and the speed with which false or manipulated material can outrun scrutiny and correction.
The legal challenge is therefore not simply to ensure that voters can physically cast a ballot. It is to protect an information environment that remains sufficiently open, plural, intelligible and contestable to ensure deliberative democracy – without converting the protection of elections into a general power to police political truth.
âTurning down the noiseâ in the digital environment
This is where the new radical idea of âturning down the noiseâ or âprotection from noiseâ is useful. The phrase should be handled carefully. It does not describe a right to truth, silence, comfort or protection from disagreement or offence. Democratic politics is ânoisyâ by design. It includes exaggeration, error, satire, anger, mobilisation and sharp contestation.
The concern is narrower, namely organised, synthetic and amplified noise. That means engineered saturation, impersonation, microtargeting, synthetic deception, covert amplification, harassment or suppression of correction in ways that materially impair the practical ability to receive, test and contest political information autonomously.
In his separate concurring opinion in Bradshaw, Judge Jakab gives this idea one of its first significant judicial formulations within the Strasbourg system. He suggests that digital disinformation can reach a point where the ânoiseâ level is so high that the right to receive information under Article 10 may itself be impaired. His concurring opinion fuses and balances human rights, platform governance, and national security considerations, at both high and granular levels of analysis.
One of the strengths of liberal democracies when compared to autocracies is that liberal democraciesâ political decisions and democratic accountability mechanisms (such as elections) tend to be relatively more strongly influenced by fact-based discourses weighing arguments. If we allow the digital disinformation noise levels to ravage as high as they currently are in our public debates, then we are endangering liberal democraciesâ structural superiority.
This may become an important line of jurisprudential development because it sits precisely where law, politics and technology now meet. There is also a historical reversal here. The post-1989 European human-rights order was significantly concerned with supporting transitions towards democracy and the rule of law. It must now consider how established democracies should be protected when revanchist authoritarian states and resurgent anti-democratic movements exploit democratic openness, sometimes in mutually reinforcing ways, to destabilise political institutions and corrode public trust from within. It must also confront the reflexive danger that the language of democratic protection may itself be exploited. Real, exaggerated, fabricated or unverified claims of malign interference may be weaponised to delegitimise elections, discredit political opponents and independent institutions, justify coercive or exclusionary measures, and consolidate power.
Giving citizens more control over the noise
The point is not that Strasbourg should prescribe a single model of platform regulation. Bradshaw itself points the other way in that in an area of technical complexity, national-security sensitivity and limited European consensus, States retain a wide margin of appreciation. But subsidiarity does not mean that human-rights law has nothing to say. As Robert Spano has argued in a different context, the contemporary Convention system increasingly works through process-based supervisory review whereby national authorities are given space where they act in good faith, reason carefully and balance competing interests, but law still helps sustain the democratic character of political decision-making. The question for future cases is whether emerging regulatory practices concerning provenance, amplification, synthetic content, political advertising and platform risk assessment will give courts firmer ground on which to assess the informational conditions of democratic choice, particularly where those conditions are deliberately manipulated to undermine the orientation towards âeffective political democracyâ identified in the Preamble to the European Convention of Human Rights.
In practical terms, protection from noise would not mean that States or platforms decide political truth for citizens. It would mean giving citizens more meaningful control over the conditions in which political information reaches them. Users should be able to know whether they are dealing with a real person, a pseudonymous account, a bot, a political advertiser, an influencer, a campaign organisation or a verified media source. They should be able to reduce the volume of automated, recommended or paid-for material; choose non-personalised or following-only feeds; see when political content is synthetic or AI-generated; and understand why they are being targeted. The aim is not to remove disagreement from democratic life, but to make the architecture of online political communication more transparent, adjustable and contestable.
2025 Romanian presidential election â noise control or authoritarian regression?
The Romanian election litigation shows the stakes. In CÄlin Georgescu v Romania, the Strasbourg Court rejected the application concerning the annulment of Romaniaâs 2024 presidential electoral process, principally because Article 3 of Protocol No 1 does not normally apply to presidential elections unless the office has sufficient legislative functions. That means Georgescu should not be treated as a Strasbourg merits ruling approving the annulment. Its importance is different. It records the kind of factual scenario now pressing against electoral law, namely algorithmic amplification, digital technologies and artificial intelligence, unlabelled electoral promotion, campaign-finance opacity and unequal campaign visibility.
The Romanian Constitutional Courtâs Ruling No 32 of 6 December 2024 treated the electoral process as vitiated by non-transparent digital promotion, misuse of social-media algorithms, artificial intelligence, and undeclared campaign funding. Whether one sees that decision as necessary constitutional safeguarding or a dangerous authoritarian regression, it illustrates the outer edge of protection from noise. At some point, information-environment manipulation may be argued to affect not merely campaign fairness, but the free formation and expression of democratic choice.
That is why the Venice Commissionâs urgent report on the cancellation of election results is so important. It recognises that online campaigning, disinformation, artificial intelligence and external influence create new electoral risks. But it also insists on constitutional discipline. Generally, annulment must be exceptional, it must rest on clear legal competence, fair procedure, adequate evidence, and a reasoned assessment that irregularities may have affected the result. In other words, democratic systems must be able to recognise information-environment harm without making election cancellation too easy.
The emerging regulatory picture is therefore fragmented but significant. Bradshaw gives the human-rights law âseedingâ that electoral rights may require positive protection against hostile interference. The Romanian litigation gives the factual pressure point, namely platform amplification and digital opacity can put electoral legitimacy under severe strain. The Venice Commission supplies the rule-of-law brake in policy terms that serious remedies require legality, evidence, procedure and proportionality.
Retreat from platform self-governance in the EU and UK
Platform governance is the next part of the picture. The EU Digital Services Act moves beyond a purely passive view of platforms. It treats very large online platforms and search engines as potential sources of systemic risk, including risks to fundamental rights, civic discourse, public security and electoral processes. The Commissionâs electoral-process guidelines under the DSA translate that idea into election-specific concerns including recommender systems, generative AI, political advertising, coordinated manipulation, crisis-response mechanisms and cooperation with electoral authorities.
The EU Regulation on the transparency and targeting of political advertising adds another layer. Political advertising becomes democratically dangerous when voters cannot tell who paid for a message, why they are seeing it, which audience was targeted, or whether different voters are receiving incompatible versions of a campaign. Transparency does not solve manipulation, but it helps restore the public character of political contestation.
Privacy and data protection also matter. Profiling and political microtargeting can transform democratic communication into the behavioural sorting of citizens and the tailored manipulation of their choices. The issue is not merely privacy in the narrow sense. It is whether voters are placed into opaque political environments selected according to inferred fear, resentment, identity or susceptibility. In that respect, the GDPR is also part of the democratic-resilience architecture of mitigating manipulation and coercion.
The EU AI Act is especially relevant because it recognises that harm may arise from system design, not only unlawful content. Its rules on manipulative or deceptive techniques, vulnerability exploitation, synthetic-content transparency and AI systems used to influence elections or voting behaviour do not create a general law of malign influence. But they identify risk indicators that map closely onto protection from noise, namely opacity, synthetic realism, behavioural steering, vulnerability targeting and loss of provenance.
In contrast, the UK relies on a dispersed framework comprising the Online Safety Act 2023, data-protection law, electoral law, including campaign-source transparency requirements under the Elections Act 2022, and the National Security Act 2023. The latter criminalises certain coercive, deceptive or otherwise unlawful foreign interference with political processes, public functions and UK interests, while the Foreign Influence Registration Scheme requires greater transparency concerning political influence conducted at the direction of foreign powers.
European regulatory leadership of noise control in the digital sphere
This is where a limited âBrussels effectâ may matter. The point is not that the EU model will become global law, nor that it is beyond criticism. But large platforms, AI providers and advertising intermediaries may find it easier to build systems around demanding expectations of transparency, provenance, audit, risk assessment and researcher access. If so, EU regulation may indirectly raise the baseline for democratic information governance elsewhere, including within the broader European legal space. That hope depends on enforcement, judicial oversight, platform incentives and whether other jurisdictions view systemic-risk regulation as democratic infrastructure rather than censorship.
The importance of Bradshaw is therefore not that Strasbourg has created a settled right to protection from disinformation. The Court recognised the possible relevance of positive obligations but, rather than prescribing particular safeguards or applying strict review, left States considerable discretion within a wide margin of appreciation. That caution reflects the absence of a clear European consensus on the measures required to protect elections from hostile information operations. It also makes it more important to identify where regulatory convergence, and potentially such a consensus, is beginning to emerge. Across human-rights law, electoral law, platform regulation, data protection, AI governance and Venice Commission standards, recurring principles include provenance, transparency, recommender accountability, limits on microtargeting, synthetic-media labelling, campaign-finance visibility, platform-data access, fair procedure and evidence-based remedies.
None of this means that technology can solve social, political, economic and diplomatic problems. Democracies cannot outsource electoral integrity to labels, dashboards, audits or content-moderation tools. But the opposite error is equally dangerous, namely leaving platforms, AI systems and advertising infrastructures without legal guidance, free to operate in a manufactured aterritorial space, immune from accountability for the destabilizing effects occurring through their digital architectures while expecting voters, journalists, candidates and electoral bodies to absorb the forced choices and consequences.
An emerging consensus on noise control?
The emerging consensus, if there is one, is not a consensus on truth control. Nor is it yet a fully settled Convention-law consensus on exactly what every State must do. There remains regulatory divergence, and much will depend on domestic implementation, enforcement capacity and institutional design. But the legal and policy conditions for convergence are now visible. Across platform regulation, electoral law, data protection, AI governance, political-advertising transparency and Venice Commission standards, recurring concerns can be identified, namely provenance, transparency, recommender accountability, microtargeting limits, synthetic-media labelling, campaign-finance visibility, platform data access, fair procedure and evidence-based remedies.
Protection from noise is best understood in that restrained sense. It is not protection from disagreement, nor a licence for State truth control. It is protection of the practical conditions that allow disagreement to remain visible, intelligible, contestable and democratic. Citizens should be able to know who is speaking, who is paying, whether content is synthetic, why they are being targeted, how platforms amplify political material, and what remedies exist when those systems are abused. The deeper danger is not only that citizens may encounter false claims, but that digital architectures may increasingly shape attention, visibility and behaviour in ways that are difficult for citizens to perceive, contest or resist. In that respect, Shoshana Zuboffâs account of âinstrumentarian powerâ is useful. Malign actors, foreign, domestic, may exploit these aterritorial architectures not by physically silencing or controlling citizens through totalitarian force, but by steering the conditions under which they see, feel, rank, trust and act in a way that is detrimental at individual and societal levels. It is to preserve enough autonomy, transparency and contestability for us to be more confident that what we take ourselves to think is genuinely our own.