Author : Ishita Agrawal from Dr. Bhimrao Ambedkar law university, Jaipur (Rajasthan) Co- Author : Prashasti Bhatnagar from Christ University, Bangalore
I. INTRODUCTION
Digital intermediaries have emerged as the primary entry doors to access information, partake in political discourse and avail rights provided under Article 19(1)(a) of the Constitution of India. But their governance mechanisms, including content moderation algorithms, recommendation engines, shadow ban and demonetisation systems have escaped significant level of transparency and accountability. In contrast to traditional censorship, which has been addressed by constitutional jurisprudence which arose around the State action doctrine, this algorithmic censorship is enforced by private corporations and not actors of government who have an obligation to respect rights falling within the ambit of the Constitution.
Reconceptualising free speech for the digital age: Reassessing the scale and opacity of algorithmic governance. This paper argues for a recalibration of free speech principles to account for the unprecedented scale at which algorithmic governance is being deployed. Private platforms cannot merely be considered as State actors simpliciter, yet their role as gatekeepers of public discourse compels that principles consonant with constitutional values publicity, accountability and shielding democratic participation are indirectly delivered to them, through a process this paper terms constitutional value infusion. The applicable constitutional framework and judicial journey, highlights regulatory challenges of the current intermediary regime, considers comparative approaches, and advances a constitutionally grounded model for algorithmic accountability in India.
II. FROM STATE CENSORSHIP TO ALGORITHMIC GOVERNANCE
Since censorship for the most part was an act of the state, constitutional doctrine grew around it. This architecture has been fundamentally disrupted by the digital era. By then, content targeting audiences and those that do not are now decided by automated recommendation and moderation systems offered by platforms like Facebook, YouTube, Instagram and X. The one thing that makes this kind of censorship constitutionally unique is invisibility, because in a great many cases the user whose post has been algorithmically demoted or completely deleted from recommendation feeds will never even know any restriction took place much less have an avenue to seek relief.
Far from simply the censorship of content, algorithmic control over visibility is also a constitutional problem. In 2024, Meta announced updates to its political content recommendations across applications including Instagram and Threads in which the company made alterations to prioritize fewer people being recommended political posts by default. There are other accounts of similar behaviour by these automated misinformation-detection systems removing legitimate journalistic and academic content without even hinting as to why the removal was warranted. In a world where algorithms decide the political speech billions of people actually see, it is a new order of communicative power one with no parallel in the analogue history of free speech.
III. CONSTITUTIONAL FRAMEWORK AND JUDICIAL EVOLUTION
Article 19(1)(a) guarantees every citizen the right to freedom of speech and expression. The Supreme Court has consistently held that freedom of speech is foundational to democratic governance and informed public participation. Article 19(2), however, permits reasonable restrictions on specific grounds public order, decency, morality, sovereignty, and the security of the State.
The constitutional difficulty posed by algorithmic censorship comes from the fact that free speech jurisprudence was largely built around State action. Four decisions, read together, chart the trajectory of constitutional doctrine into the digital space.
In Shreya Singhal v. Union of India (2015), the Court struck down Section 66A of the Information Technology Act as unconstitutionally vague, distinguishing between discussion, advocacy, and incitement, and affirming that unpopular or controversial speech is constitutionally protected. The chilling effect produced by opaque and unpredictable algorithmic enforcement is structurally identical to that of vague penal legislation: in both cases speakers self-censor because the rules of engagement are unknowable.
In Anuradha Bhasin v. Union of India (2020), the Court has ruled that freedom of speech and expression includes access to the internet and access restrictions must meet the standards of proportionality. The principles of meaningful participation in digital space carry over readily: meaningful participation in the name only is participation that is not meaningful.
Justice K.S. Puttaswamy v. Union of India (2017) has held that informational autonomy, or the right to control informational environment, is a value to be protected by the State. This dimension of the right of privacy has a direct bearing where algorithms have an influence over political awareness and electoral discourse.
Last but not least, in Kaushal Kishor v. State of Uttar Pradesh (2023), the Court again held that the State has a positive obligation to respect Fundamental rights even in a private sphere context where it is deemed that the private sphere possesses a sufficient degree of private power. Together, these decisions suggest a judicial sensibility: constitutional freedom exists in the digital realm, and it is not enough to access the machinery of the constitution if the exercise is meaningless, constitutional values do not lose their vigor merely because the threatening entity is a corporation and not the State.
IV. THE CONSTITUTIONAL DILEMMA OF PRIVATE PLATFORMS
A. The Article 12 Path Constraints
Some scholars point to the fact that dominant social media platforms undertake public functions and are subject to a regulatory framework created by law, and argue that they should be considered as ‘State’ under Article 12. The argument is blocked by insuperable doctrinal obstacles. In Pradeep Kumar Biswas v. Indian Institute of Chemical Biology (2002) the Supreme Court held that the inclusion of the term “State” in Article 12 requires both functional and structural control by the government, a threshold that private technology companies, however large, do not meet. More importantly, this route is not necessary. A more manageable route is through the indirect application of constitutional values through legislation and common law.
Yet platforms like Meta, Google, YouTube, and X increasingly resemble public communication infrastructure. Their influence over speech often dwarfs that of traditional media organisations. The United States Supreme Court recognised this reality in Packingham v. North Carolina (2017), describing social media platforms as the modern public square where citizens engage in political and social discourse.
B. Infused Constitutional Values:
There’s a model of constitutional value infusion: the legislature and courts prescribe that transparency, accountability, due process and non-discrimination are to be considered design principles for platforms that wield communicative power in a dominant way but are not state actors and, therefore, not subject to constitutional restrictions. This is based on the reasoning of Kaushal Kishor and the general principle that constitutional values are reflected in the whole legal order, like the German Federal Constitutional Court’s doctrine of Mittelbare Drittwirkung (indirect horizontal effect), which holds that constitutional values extend into private law relationships, usually where there is a marked power imbalance.
The counter-argument of editorial autonomy on the platforms that algorithmic curation is protected expressive conduct is credible but not adequate at the scale implicated. A newspaper’s editorial discretion is not similar to the systematic suppression of political speech by a near-monopolistic platform across hundreds of millions of users. Scale alters the constitutional dynamics. The counter-argument which transparency requirements may lead to government intervention is the stronger one. The solution is institutional: oversight must be placed on truly independent institutions that are largely free from directives from the Executive and that examine process and systemic risk instead of specific content issues.
C. Constitutional Value Infusion: The Better Framework
A model of constitutional value infusion is argued based on the logic of Kaushal Kishor and the principle that constitutional values infuse throughout the legal order. In this model, the legislature and courts equate transparency, accountability, due process and non-discrimination as necessary design features for platforms that have dominant communicative power, but are not “state actors. This is by no means a new concept. The Doctrine of Mittelbare Drittwirkung (indirect horizontal effect) of the German Federal Constitutional Court has long recognized that certain constitutional principles and principles extend into private law relationships, where a significant power imbalance exists. The Indian position is less developed, but it is in a direction which is compatible. By the same logic the State has an affirmative responsibility to establish circumstances in which fundamental rights can be exercised meaningfully, as the Supreme Court said in PUCL v. Union of India (1997) and that responsibility applies to regulating private actors who pose a threat to those circumstances.
V. REGULATORY GAPS IN INDIA
At present, the Indian law governing digital intermediaries is the Information Technology Act, 2000, and the IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021. Safe harbour protection is offered in Section 79 as per due diligence requirements; content removal and grievance redressal requirements are included in the Rules. The system, however, doesn’t touch the problem of algorithmic decision making, the unlawful content. There is no comprehensive understanding of the obligations that platforms have to reveal their content ranking practices, clarify to the user why they have made a particular decision about content, allow external audits of their content recommendations systems, or make open and transparent the systemic operation of algorithmic governance.
In addition, there’s a structural problem: the Intermediary Rules were devised and managed by the Ministry of Electronics and Information Technology, precisely the very same executive part that could want to regulate political speech. Both the Digital Personal Data Protection Act, 2023 and the Draft Digital India Act, 2023 are part of a vision to reform digital governance more broadly, but neither of them explicitly mentions algorithmic content governance or the concerns of the Constitution regarding visibility control. The gap in regulation remains.
VI. COMPARATIVE PERSPECTIVES
The EU’s Digital Services Act (DSA) requires very large online platforms to be transparent and assess risks, forcing them to reveal their content moderation practices and algorithmic systems, and to offer at least one recommendation option that does not use user profiling. In contrast, the US remains fragmented, with broad platform immunity still afforded under Section 230 of the Communications Decency Act and the Supreme Court’s unwillingness to address the constitutional nature of algorithmic curation in Moody v. NetChoice (2024), suggesting deep uncertainty. China’s Provisions on the Management of Algorithmic Recommendations (2022) makes algorithmic governance subordinate to State ideological control and ensures compliance with governmental oversight, exactly the risk an Indian framework must be designed to avoid.
A clear spectrum is emerging: EU prioritises transparency based on rights, US prioritises autonomy of platforms, China prioritises state control. India does not consistently espouse any of these positions. The DSA model is the closest analogue to India’s needs, but in view of the concerns identified in Part V above, would need to be modified to insulate the regulatory process from executive direction.
VII. TOWARDS A CONSTITUTIONAL MODEL OF ALGORITHMIC ACCOUNTABILITY
The following principles are constitutional and should be implemented in the future by regulation.
Transparency reports with information on the nature, scale and basis of content moderation and algorithmic ranking decisions should be obligatory for platforms that reach a prescribed level of number of users. When algorithmic actions have a significant impact on the visibility of the user’s content, users must be communicated in an understandable manner. Non-compliance should be sanctioned through graduated penalties, and if someone repeats the offence, there is a risk of losing the safe harbour provision under Section 79.
Second, a statutory right to explanation should be added. The person whose content is deleted, restricted, or demonetised will be supplied with a meaningful statement of reasons, a process that appeals, and a qualified and impartial reviewer. The standard should be similar to the principles of natural justice which are already entrenched in Indian administrative law, notice, hearing, and reasoned decision.
Thirdly, the top platforms should be subject to independent algorithmic audits. These audits would assess systemic risks concerning political bias and discrimination and the arbitrary restrictions of lawful speech. Penalties should be limited proportionate to the amount of revenue that the platform gets from India so that there is no disproportion between the penalties and the size of the platform.
Fourth, India should set up an independent Digital Communications Ombudsman with statutory powers to hear and review major content moderation issues, to conduct serious investigations on complaints and provide binding recommendations. The Ombudsman shall be structurally independent of the executive and appointed by a process similar to that for appointment of constitutional office-holders; and shall be removable for reasons similar to those for removal of a constitutional office-holder.
Fifth, and more fundamentally, legislation should explicitly acknowledge that “algorithmically governed” governance of dominant online platforms must be done in ways that respect the constitutional values of freedom of expression, informational autonomy, and democratic participation. It is not the same as making platforms constitutional duty-bearers that it will be using the power of legislative edicts to direct constitutional norms to private relations in just the way that Kaushal Kishor leaves open. If an algorithmic design of a platform is consistently inconsistent with these values as a result of an audit or investigation by an Ombudsman, the statutory statement of the constitutional obligation should set the basis for a court to subject the design of the platform, not only that of individual moderation decisions, to constitutional review.
VIII. CONCLUSION
The constitutionality of algorithmic censorship is one of the biggest challenges of this digital era. Some technology firms, in a way and in a degree that is unprecedented, are wielders of communicative power; they determine what political speech billions of people can hear, by means of automated systems that are largely invisible and unaccountable. The architecture of Indian constitutional law which ensures protection to the individual against the State and leaves private actors wide open is in built-out form structurally unsuitable for this reality.
No, the answer is not to add Article 12 to cover private platforms, because doing so would not work doctrinally and would be a new threat for the State to have to deal with in terms of content governance. Nor is it sufficient to say ‘let others rule their own platforms as they please’ when platforms are the public as they are now, but when they were a marginal way to rule their own.
The constitutional value infusion through independent regulation taken by the model put forth here lies somewhere in between the two extremes of constitutionalism that Indian constitutional jurisprudence already seems to be heading toward. Shreya Singhal says that speech suppression should be clear and fair. Access to digital spaces is part and parcel of constitutional freedom, says Anuradha Bhasin. According to Puttaswamy, informational autonomy is a constitutional value which the State is obliged to safeguard. And Kaushal Kishor says that principles of constitution do not end at the line of public and private. The freedom guaranteed by the part of Article 19(1)(a) goes beyond freedom to post, it is the freedom to meaningfully engage in democratic conversation. That freedom will require the design, governance and accountability of private systems in the algorithmic era. Indian constitutional law has the means to tackle this. It’s not that the legal and legislative imagination is not there to use them, it’s not.