Author(s): Aditya Gurjar
Paper Details: Volume 4, Issue 3
Citation: IJLSSS 4(3) 08
Page No: 87 – 97
ABSTRACT
The swift expansion of AI technologies has brought about significant changes to digital platforms, moving them beyond simple information intermediaries to actual participants in content creation. The established legal doctrines concerning intermediary liability were originally grounded in the understanding that intermediaries merely hosted and transmitted the content of their users and that they do not exercise any kind of editorial control over it. However, this paradigm has clearly changed due to the emergence of AI systems capable of creating text, image and video in an independent manner. In this paper, we scrutinize how relevant legal doctrines – intermediary immunity regulations – remain in the context of an AI generated environment. A critical assessment through the doctrines of laws and comparisons between India, the European Union and United States has examined the change of status of platforms, issues concerning assigning responsibility, and the resulting consequences on rights like free expression and privacy. The arguments here support a restructuring of the status of platforms and the introduction of a mixed framework for regulation, which is balanced between fostering innovation and guaranteeing responsibility. Ultimately, we have claimed that a new legal paradigm must be established to provide a status to the platform not as passive entities but active participants in the production.
Keywords:- AI-Generated Content, Intermediary Liability, Platform Accountability, Algorithmic Governance, Digital Regulation
INTRODUCTION
In recent years, the digital landscape has been significantly transformed by the rise of artificial intelligence and machine learning technologies. Where the internet’s intermediaries once provided a passive platform for users to share and access information, many are now actively shaping the way content is created, curated, and distributed. Technologies such as generative AI (capable of creating highly realistic text, images and videos) can generate vast amounts of sophisticated content, often indistinguishable from content generated by users. This development undermines the traditional liability rules governing intermediary liability, and the scale and speed at which AI can create content means both the reach and impact of digital communication is now magnified. Furthermore, the ease with which these powerful tools can be used and applied to platforms makes the distinction between content created by humans and that generated by algorithms very difficult to discern. This poses challenging questions concerning control, ownership, and responsibility. As such, law is faced with a pressing need to revisit its underlying assumptions about digital regulation.
For years, the law of intermediary liability has protected platforms against liability for content uploaded by third parties as long as they have acted as simple conduits (or passive intermediaries), through the ‘safe harbour’ provisions designed to encourage the free flow of information and development of the online sector.[1] However, the rise of AI-generated content raises concerns in this regard, as it is the intermediary platform that now plays an active role in creating and disseminating content, without direct human input.
The present paper attempts to explore the following: are platforms that have deployed generative AI, able to claim they are passive intermediaries, or are they new types of active creators? Examining the trajectory of intermediary liability, the powers of the AI systems and the limitations of the current framework, this paper attempts to suggest a reformist model to platform liability.
RESEARCH PROBLEM
The advent and quick assimilation of artificial intelligence to digital platforms has shattered the basic premises underpinning intermediary liability frameworks in different jurisdictions. The doctrines governing intermediary liability have been historically developed on the assumption that platforms are inert conduits facilitating or transferring content generated by third parties without exercising any editorial discretion over said content, therefore, enabling the safe harbour exemptions where intermediaries cannot be held liable for third party content. Nevertheless, generative AI technologies has fundamentally changed this equation by allowing platforms to not only host but also generate and promote content through the algorithmic processes in the intermediary systems.
This change is what introduces a significant problem for legal frameworks and principles. It is becoming apparent that the current immunity doctrine for intermediary platforms does not reconcile well with the affirmative actions performed by platforms operating within AI driven content ecosystems. When an AI technology developed or implemented on a platform is employed for autonomously generating a text, image or video, it blurs the boundaries between user generated content and platform generated content. In such context, platforms cannot claim safe harbor status based on the justification that they are simply passive intermediaries since they now become an integral player in the creation and publishing of content through the algorithms.
These challenges have implications regarding attribution, accountability, and transparency. The “black box” nature of AI systems does not allow the traceability of the decision making processes behind them; and this consequently makes it difficult to hold them accountable for issues such as defamation, libel, privacy violations and misinformation. Besides, the absence of well-established legal standards governing AI-generated content adds up to the current level of uncertainty for platforms, users, and regulators alike.
In essence, the main problem is to reconcile intermediary liability doctrines to the current reality brought by AI technologies applied in intermediary platforms and establish if a new regime should be implemented.
RESEARCH OBJECTIVES
This research aims to conduct a thorough analysis of how the functionalities of the digital platforms evolved from being intermediaries to content generators under the impact of AI technology, and whether the current regime on intermediary liability still applies with regard to issues raised by the production of content using AI, how the safe harbour mechanisms of today’s world stand with changing intermediation structures and where and how do the existing laws fall short.
It also intends to undertake a critical analysis of assigning liability in the production of content using AI with regard to matters of accountability, transparency and algorithmic decision making. The research would conduct a comparative study of different regimes, specifically India, European Union and United States, in order to highlight and ascertain the response of different legal regimes to the problems at hand.
It would further analyze the impact of platform liability on human rights such as freedom of expression and right to privacy. Lastly it intends to formulate a model which can serve as an effective alternative legal framework in this regard that balances innovation with responsibility.
RESEARCH QUESTIONS
- Can digital platforms that deploy generative AI technologies still be classified as passive intermediaries, or should they be legally recognized as active creators of content?
- To what extent are existing intermediary liability frameworks (such as safe harbour provisions) adequate in addressing the legal challenges posed by AI-generated content?
- What regulatory model can effectively balance platform accountability with the protection of fundamental rights like freedom of expression and privacy in the age of AI-driven content creation?
RESEARCH METHODOLOGY
The research follows a doctrinal and analytical method to critically examine the shifting landscape of platform liability in light of AI-generated content. The study uses primarily secondary sources of data which consist of statutes, case laws, academic writings, policy papers and international regulatory instruments. Several legal principles concerning intermediaries liability will be explored for the purpose of understanding their reach, meaning and limitations in dealing with the concerns emerging out of generative AI.
A comparative legal method is also undertaken for the examination of the intermediary immunity and platform liability in various countries like India, European Union and United States. Through the use of this approach, commonalities, differences and good practices across jurisdictions can be understood.
Furthermore, a critical analytical method is used to assess the adequacy of existing legal doctrines in the face of developing technologies. The shift from passive intermediary to active content creator and its impact on the distribution of liability, accountability and regulation is analyzed. A multi- disciplinary approach taking a cue from technology studies and ethics is also engaged for the understanding of AI systems. The study also incorporates the constitutional aspects of the regulation of intermediary including freedom of expression and privacy. Through this approach the study seeks to provide an holistic understanding of the legal challenges of AI generated content and a balanced and futuristic framework.
EVOLUTION OF INTERMEDIARY LIABILITY- FROM NO LIABILITY TOWARDS LIABILITY
The legal regime on intermediary liability grew out of the dawn of the internet, as legislatures sought to strike a balance between the need for accountability, and the necessity to encourage innovation in the domain of technology. Earlier regimes recognized that if platforms are made to be responsible for every piece of user generated content then the innovation would suffer; consequently a lot of jurisdictions developed safe harbour provision to exempt intermediaries from liability under certain conditions.
In India section 79 of the IT Act 2000 accords immunity to the intermediary as long as they do not initiate the transmission, select the receiver of the transmission, and do not modify the information in the transmission.[2] Similarly in the US section 230 of the Communications Decency Act offers broad immunity to the intermediaries, where it allows them to host the user generated content without treating them as a publisher, whereas the EU, uses an intermediary liability approach with a two prong attack, on one hand under various instruments including E-commerce directive, and recently the digital services act, intermediaries will be imposed with various due diligence obligations, and on other hand, immunity may still be afforded under the specified circumstances.[3] The assumption is that intermediary is only passive. However in the case of AI enabled platforms this assumption of passive intermediary has ceased to hold.
Over time, intermediary liability framework has evolved much before the dawn of generative AI. Intermediaries have been pushed to take actions such as content moderation, algorithmic ranking of the content and targeting advertising thereby exercising some form of editorial control. Various interpretations by the judiciary further added on to this evolving regime. Certain jurisdictions restricted intermediary immunity where platforms are found to be privy of unlawful content, failed to act thereupon, or where the platform itself contributed towards such transmission. This evolution saw a gradual movement from non-liability toward conditional liability in accordance with the actual and knowledge.
This coupled with the increase in the usage of social media and spread of the digital ecosystems has intensified the concern regarding various social problems including misinformation, hate speech etc. This is resulting in a trend where intermediaries are increasingly being forced to carry out pro-active actions, including content takedowns, grievance redressal mechanism and transparency norms etc. This only shows that the intermediation is gaining strength from being just a tool to something much larger in the form of ‘public sphere intermediary’.
The AI functionality will carry this evolution further. Unlike earlier technologies, the AI system itself can not only process, organize or transmit information but also contribute to its creation, usually without the intervention of the human mind. This is the fundamental shift which challenges the rationale of intermediary liability framework. In fact, when intermediary starts creating the content itself it may become rather difficult to call them a mere platform.
In this evolving framework it is crucial to reiterate the trend and to try to understand that the intermediary liability has shifted from no liability towards some form of conditional responsibility, and in the given scenario it can lead towards an enhanced intermediary liability regime.
THE ERA OF GENERATIVE AI AND THE SHIFT IN PLATFORM CHARACTERISTICS
Generative AI brings about a revolution in content creation and consumption on platforms. Instead of solely relying on user interaction, generative AI systems have the capacity to generate their own content through algorithms and training datasets. This includes, but is not limited to, the use of AI systems in chatbots, recommendation systems, and content generators, all of which are now integrated within platforms. Such rapid developments have exponentially increased the volume, speed, and complexity of online content creation. AI systems can accurately mimic not only human language and artistic styles but also behavior patterns, which has led to greater reliance on these autonomous systems for communication, customer service, and information sharing. It is for this reason that platforms no longer rely on human intervention for content generation, and AI autonomy challenges traditional notions of authorship and platform control.
The integration of AI into platforms changes their role. Platforms are no longer solely facilitators; rather, they become active participants in content generation and dissemination. It is within the control of AI algorithms as to what information will be delivered to users, the responses provided for user queries and even completely original content. In this regard, questions arise as to how current intermediary liability rules would apply to this type of situation. The boundaries between platform-generated and user-generated content becomes blurry. If an AI system generates misleading content, to whom is liability attributed? Is it the user, the platform or the AI developers?
DIFFICULTIES OF APPLYING ESTABLISHED INTERMEDIARY LIABILITY FRAMEWORKS TO AI GENERATED CONTENT
The challenges of applying traditional intermediary liability frameworks to AI generated content can be stated in two points. The concept of passivity is contradictory to the active role that AI technology plays in content generation. By enabling AI technologies platforms can not rely on the fact that they are not directly controlling content, since these platforms also own and design the relevant algorithms that control the output of the AI generated content. From the perspective of content, it can thus be said that the platforms are not passively relaying content but actively creating it. Traditional interpretations that platforms are neutral intermediaries thus lose justification. The threshold of immunity for intermediary liability, based onnon-interference, seems to be an antiquated approach. Thus there exists a gap between technology and the law. Secondly there lies an issue of attributing the content. Since AI technologies work autonomously, determining a responsible party and attributing the responsibility in cases of problematic AI generated content is more difficult. Who is liable? The platform, the programmers of the AI, the end-user? The opacity of AI technologies (known as the ‘black box’ problem) further complicates the ability of an entity to be assigned liability. The reasons that an AI generates specific output cannot always be easily determined. Lastly, AI generated content can be produced in extremely high volumes and at remarkable speed, which in effect causes and disseminates harm at a pace that cannot be controlled. As the content is virally spread across the internet by technology it causes extensive damage long before measures are in place to control it, placing considerable burdens on regulators to identify the source and to impose some sort of enforcement, as well as on platforms to ensure compliance. Notice and takedown systems may not be sufficient when harm occurs almost instantly.
COMPARATIVE LEGAL ANALYSIS
A comparative examination of the legal frameworks of these countries illustrates divergent approaches to tackle the complex issues brought forth by AI-generated content. In the United States, the expansive immunity given to platforms under Section 230 of the Communications Decency Act, has increasingly come under fire as critics have sought reforms aimed at addressing the platform’s role in moderating and generating content.[4] The proposed legislative changes, however, have been few and far between while the status quo under the current legal doctrine continues to extend generous immunity even as platforms are actively involved in curating and producing content. This dissonance between the state of technology and legal frameworks poses a critical tension, with the legal regime not adequately addressing the harms derived from algorithms and AI generated content, yet concerns persisting about stifling technological innovation and freedom of speech. Hence, the United States has not been able to establish the right equilibrium between regulatory interventions and technological freedom.
In contrast, the Digital Services Act in the EU provides a forward-thinking framework for regulation. It places obligations on the platforms for risk assessments and mitigating systemic risks arising out of algorithms and does not explicitly mention generative AI but gives a foundational framework for regulating AI-driven platforms through the mandates of transparency and accountability.[5]
In India, the law is in its nascent stage, with new Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 introducing more responsibility and due diligence obligations on intermediaries. Nevertheless, they don’t specifically cover generative AI related issues.[6]
This comparative study demonstrates that a much comprehensive and coherent framework for platform liability considering the capacities of AI technologies is needed.
CONSTITUTIONAL AND RIGHTS-BASED ISSUES
Questions regarding freedom of expression and privacy are crucial in light of regulating AI-generated content. Firstly, imposing platform liability can promote over-censorship, wherein platforms would actively try to evade liability by eliminating all controversial content, which would lead to stifling free speech. Any such cautionary removals might impact non-infringing speech, dissent and minority opinions more severely, and in the case of an automated or overbroad system, the infringement may become greater. In such a situation, the fine line between regulating and not curtailing free speech would become much finer and it would depend greatly on the extent and way the AI is regulated.
Secondly, non-regulation of such content can cause irreparable harm to the individual and society, including defamation, dissemination of fake news and violation of privacy. The balance between both aspects seems to be challenging.
In the Indian legal framework, freedom of speech and expression under Article 19(1)(a) is subject to reasonable restrictions under Article 19(2). While implementing the regulations on AI-generated content, the Indian courts would probably apply the principle of proportionality established in case law regarding restriction of fundamental rights. If the restriction on free speech by such regulations becomes excessive and unreasonable, it would likely be struck down as violating Article 19(1)(a). Similarly, privacy has been deemed a fundamental right under Article 21 in Justice K.S. Puttaswamy (Retd.) v. Union of India and any regulation or usage by the AI must conform to this right.[7] The judgement establishes that the right to privacy is intrinsic to the right to life and personal liberty, and that this would include the right to be free from such unauthorized use, analysis and dissemination of personal information; this protection also covers such inference of private information which could not have been inferred through less intrusive means. The potential to infer such information is increased exponentially by the capabilities of an AI.
REVISING PLATFORM LIABILITY: A NOVEL APPROACH
To navigate these obstacles, an approach to platform liability that considers the active role played by platforms in the context of AI, while maintaining the innovation imperative, is vital. The framework of liability presented in this paper is one that moves beyond the distinction between intermediaries and publishers to a more functional approach based on the involvement in content creation. It should incorporate the reality of technology in its application and be adaptable for the future without stifling the very innovations on which our society now relies. Hence, a flexible approach to regulatory design to tackle the challenges posed by AI would be desirable.
Firstly, platforms that utilize generative AI should be subject to increased duties of care, for example by ensuring that procedures are put in place for tracking and controlling harmful AI generated content as well as transparency of algorithmic workings.
Secondly, differentiation between types of platforms based on their role in content creation. A distinction is needed between the active generation of content and the mere hosting of it.
Thirdly, “algorithmic accountability” should be embedded in the framework of liability; it should impose an obligation on platforms to be able to explain the functioning of their AI algorithms and create avenues for redress in the event of damage being incurred.
Fourthly, a system of “distributed liability” could be established, wherein the platform, developer, and user share liability based on their contribution to the harm. This approach may alleviate the undue pressure on a single actor due to the complex nature of AI algorithms, promote greater coordination and compliance from each actor towards mitigating risk and adhering to ethical standards in its deployment. This approach contributes to a fairer and more proportionate assignment of responsibility and, critically, it allows for scenarios wherein both human and machine interaction could contribute to the damage incurred to be incorporated into the framework.
Fifthly, there is a crucial need for a harmonized approach, at an international level, for regulating AI-driven platforms.
CONCLUSION
The rise of generative AI has significantly changed the purpose of digital platforms, shaking the foundations of intermediary liability. The existing conception of digital platforms as mere passive conduits is no longer acceptable, given that these platforms actively produce and manage content, particularly by employing AI. Consequently, the established legal frameworks may no longer be suitable, and they must be updated to address the challenges posed by AI-produced content. Legal systems must evolve and abandon old classifications to adapt to a more dynamic approach. The increasing significance of AI in our content environment necessitates a careful review of how rights and obligations are allocated, since current laws may otherwise become irrelevant. The credibility of our digital governance structures relies on how accurately they correspond with the realities of technology.
This article has made the case that current intermediary liability regimes are flawed, as they are not designed for the realities of AI-influenced platforms and require redefinition. By using a hybrid regulatory model focused on accountability, transparency, and proportionality, it is possible to reach a balance between responsibility and innovation.
The law must be adaptable to the constant advancements in AI technology. One of the central difficulties is not only in developing regulations for technology but also in doing so in ways that uphold basic rights and the common good. The progression from passive conduits to active creators signifies a new stage in the development of digital law that requires a well-thought-out revision. In the future, the law must be both dynamic and accessible, and lawmakers, jurists, and industry leaders must cooperate in an effort to devise and execute effective and unified regulations for technology to encourage responsible development and progress.
[1] Shreya Singhal v. Union of India, (2015) 5 SCC 1.
[2] Information Technology Act, 2000, § 79 (India).
[3] Directive 2000/31/EC of the European Parliament and of the Council of 8 June 2000 (E-Commerce Directive); Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 (Digital Services Act).
[4] Communications Decency Act, 47 U.S.C. § 230 (1996) (United States).
[5] Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 (Digital Services Act).
[6] Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (India).
[7] Justice K.S. Puttaswamy (Retd.) v. Union of India, (2017) 10 SCC 1.