New Delhi — Meta is engaged in its second week of discussions with the Indian government, as a complex debate unfolds regarding content moderation practices on its platforms. The core of these talks revolves around critical questions concerning the extent of human intervention in reviewing content, the necessity for reviewers to comprehend Indian languages, political nuances, and local context, as well as the appropriate methods for labeling deepfakes, and the speed at which unlawful content must be removed. Another significant point of contention is whether Meta’s recommendation systems could jeopardize its status as an intermediary.
Legal experts offer diverse opinions on the government’s authority and Meta’s obligations, but they largely concur that the dispute hinges on the scope of intermediary protections when platforms actively recommend and amplify content. They are scrutinizing whether automated recommendations could effectively transform a platform into a publisher, the government’s power to mandate human review and algorithmic oversight under current regulations, and how rapid takedown requirements can be balanced with principles of free speech and due process. Concerns have also been raised about Meta’s investment in Indian-language moderation, the legal and constitutional implications of deepfake labeling, and the evolving global landscape of platform liability.
Five key legal and policy questions are at the heart of these discussions:
1. Can Meta’s recommendation algorithms affect its safe-harbor status?
The government is scrutinizing whether Meta continues to meet the requirements for intermediary protection under Indian law. Officials indicate that if platforms dictate which content is presented to users, particularly through recommendation systems or promoted paid content, this could be construed as publishing, thereby obligating them to take responsibility for such actions.
Apar Gupta, founder of the Internet Freedom Foundation, disputes this interpretation. He argues that Section 79(2)(b) of the Information Technology Act pertains to human choice in selecting a transmission’s receiver, contrasting with automated systems that apply predetermined criteria. Gupta cautions that reinterpreting automated recommendations as equivalent to content selection could strip immunity from various ranked services like search engines and app stores. He also notes that Indian law lacks a distinct "publisher" category for platforms, and losing Section 79 protection would lead to ordinary liability, not an automatic reclassification as a publisher.
Jyoti Panday, a tech-policy researcher, highlights that intermediary liability laws were designed to protect tech companies as "passive hosts." In India, safe harbor is conditional on platforms acting against specified illegal content categories within a set timeframe under Section 79. Panday suggests that if a platform’s recommendation algorithm actively curates, prioritizes, and amplifies illegal content, it risks losing its immunity. She cites US litigation involving TikTok, where courts are examining recommendation algorithms separately from third-party content, illustrating a global shift in intermediary liability regimes. The government’s examination of Meta includes whether recommendation engines and paid-content systems create responsibilities beyond those typically associated with an intermediary. The ultimate determination of a platform’s status rests with the courts.
2. Can the government mandate human review and should reviewers understand Indian languages and politics?
A significant demand from the government is for increased human involvement in content moderation. Officials argue that automated moderation often fails to differentiate genuinely harmful content from political satire, commentary, or legitimate speech within the Indian context. Consequently, the government is pushing for human reviewers who understand Indian languages, politics, and local social contexts. This is particularly crucial for sensitive cases involving prominent public figures, where the government advocates for human intervention before automated action and a minimum of two reviewers.
While Panday’s analysis suggests safe harbor was designed for passive hosts, raising concerns when systems actively curate content, Gupta states that the government can partially mandate human oversight under existing IT Rules. He points to Rule 4(4) of the IT Rules, which requires significant social media intermediaries to deploy automated tools for content like child abuse imagery, with human oversight, and Rule 4(1)(a) requiring a Chief Compliance Officer. However, Gupta argues the current framework does not mandate human review for every automated decision, nor does it empower the Ministry of Electronics and IT to demand ranking weights or source code. The government’s primary concern is that human intervention must be meaningful and contextually aware, a concern heightened by an earlier objection to action taken against a post involving Prime Minister Narendra Modi.
3. What legal duties does Meta have when content violates Indian law but not its own policy?
The government has expressed concerns about content that, while not violating Meta’s community guidelines, may raise issues under Indian law, including child sexual abuse material, content with public-order implications, and other potentially unlawful material. The government maintains that a platform’s community standards cannot solely determine compliance with Indian law, asserting that Meta must adhere to Indian law.
Officials acknowledge that not all flagged content is necessarily unlawful. Some material, such as deepfakes or political satire, may be protected under Meta’s free-speech policies. Meta has confirmed engagement with authorities, with most flagged content being removed, except for posts constituting free speech or political satire.
Gupta emphasizes Meta’s obligations of transparency and accountability to its Indian users, pointing to its underinvestment in Indian-language content review. He cites disclosures indicating that 87% of Meta’s misinformation-classification budget covered English. However, Gupta also stresses that platforms should not be solely blamed for the online information environment, as the elected government also bears responsibility. The government underscores that the issue extends beyond Meta’s internal policies to its obligations under Indian law, with the debate over safe harbor also factoring into this question, as protection is conditional on timely action against specified illegal content.
4. How should deepfakes be labeled and can the government demand stronger safeguards?
Deepfakes represent a significant concern for the government, which is scrutinizing Meta’s approach to labeling manipulated and synthetic content. Officials note partial compliance and Meta’s prioritization of prominent individuals, arguing that the problem is broader than just high-profile cases. The government is examining both technical and legal aspects, including the sufficiency of existing safeguards and the potential for independent audits of algorithms.
Gupta notes that Meta platforms already have labeling mechanisms, citing the February 2026 amendments to the IT Rules that introduced requirements for prominently labeling synthetically generated information with permanent provenance metadata. He also highlighted concerns raised by the Internet Freedom Foundation regarding compelled speech and the reliability of detection tools. The government is also considering independent audits of algorithms as part of a stronger checks-and-balances framework, with technical teams continuing discussions on safeguards, automated moderation, and identifying and labeling manipulated content.
5. Can the government require three-hour takedowns while preserving due process and free speech?
The government has reiterated its demand for faster content takedowns, including the three-hour timeline introduced in the February amendment to the IT Rules. Officials argue that rapid action is crucial due to the swift spread of unlawful content, particularly manipulated material. Meta has reportedly committed to greater due diligence in addressing unlawful content after its second round of meetings.
The government believes platforms possess the technology and resources to comply with the shorter timeline, emphasizing the increasing importance of speed. However, this raises questions about how platforms can make legally and contextually sensitive decisions within such a brief period, creating a tension between the need for speed and the demand for greater human intervention and India-specific contextual expertise. The government has confirmed that the three-hour requirement applies to specified unlawful-content takedown directions under the amended framework.
Panday views the speed requirement within the broader safe-harbor agreement, where platforms receive immunity in exchange for complying with obligations, including timely action against illegal content. Gupta argues that the government’s regulatory approach is attempting to address a wide range of platform issues through intermediary-liability provisions, raising questions about the limits of delegated regulation. The three-hour requirement is now part of the amended rules, and the government expects platforms to comply.
The demand for greater human intervention gained urgency after concerns were raised about action taken against content involving Prime Minister Narendra Modi, reinforcing the government’s view that algorithms may not always grasp the significance or context of political content in India. Joel Kaplan, Meta’s Chief Global Affairs Officer, apologized to the minister for the error restricting PM Modi’s post.
Beyond Meta, the government is exploring a broader framework and may call in other platforms to assess their compliance with Indian law and intermediary obligations. Officials emphasize that the issue transcends a dispute with a single tech company, asserting that global platforms operating in India must respect the country’s laws and constitutional framework while ensuring enforcement does not unduly restrict legitimate speech. The next phase of engagement, following global leadership-level meetings, will involve Meta’s India and technical teams, continuing discussions on moderation systems, human intervention, deepfake safeguards, and technical guardrails for automated decisions.
