Australia Sues Telegram Over Extremist Content: New Liability Test

On 30 July 2026, Australia’s eSafety commissioner, Julie Inman-Grant, announced that the country is taking Telegram to court for allegedly failing to remove “pro-terror” material, including video of the 2019 Christchurch mosque shootings and the 2022 Buffalo shooting. The lawsuit, filed under Australia’s Online Safety Act, seeks a fine of up to A$54.6 million (US$38 million). Inman-Grant accused Telegram of maintaining a “permissive environment” that made extremist content “very easy to find” and alleged that the platform was “sometimes used to plan attacks” BBC News.

Telegram disputes the claims, stating its anti-terrorism measures are “extensive” and that it will contest the case in court. The lawsuit comes just a day after Russia charged Telegram founder Pavel Durov with facilitating terrorism, and follows a protracted regulatory battle: the eSafety office had been inquiring about Telegram’s content moderation practices since March 2024, enduring “five very difficult months of unresponsiveness” before the platform began engaging. Australia previously fined Telegram A$1 million in February 2025 for failing to respond on time about its handling of child abuse and extremist material.

The Australian case is part of a broader global trend: courts and regulators are increasingly testing the limits of platform liability, moving beyond simple notice-and-takedown regimes toward holding platforms accountable for the structural design of their services. Two other recent developments—the EU Court of Justice’s Coyote System ruling and India’s updated IT Rules—illustrate this shift in distinct but complementary ways.

The EU’s Algorithmic Liability Test

In July 2026, the EU Court of Justice issued a judgment in Coyote System (Joined Cases C-188/24 and C-190/24) that could fundamentally alter how platforms claim the hosting liability exemption under the e-Commerce Directive. The court held that when an intermediary’s algorithm determines “under what conditions, how and in which order of priority” information is disseminated, the intermediary “controls” that information and loses its protection from liability Electronic Frontier Foundation.

The ruling marks a significant departure from earlier case law. In YouTube and Cyando, the Court of Justice had recognized that platforms could still benefit from the hosting exemption despite categorizing, ranking, and recommending user content through algorithms. The key question, according to the Advocate General in that case, was “intellectual control of that content”—who controls the information itself, not how it appears. Coyote System shifts the emphasis by equating algorithmic organization with content control.

The Electronic Frontier Foundation (EFF) has warned that this ruling could have “terrible consequences for freedom of expression.” Platforms may engage in proactive monitoring and removal of lawful content to avoid liability, chilling speech in ways that conflict with the Digital Services Act (DSA), which explicitly accepts recommendation algorithms as a defining feature of modern platforms and regulates them through due diligence obligations rather than stripping immunity. The EFF also notes that the Coyote System ruling is part of a pattern—other recent judgments, such as Russmedia and AGCOM, have gradually narrowed intermediary liability protections.

India’s Deepfake Due Diligence

Meanwhile, India is testing platform liability through a different route: AI-generated synthetic content. The Bombay High Court recently allowed Union Minister Nitin Gadkari to institute a civil suit against Meta, X, Google, and YouTube over alleged AI-generated deepfakes. The case brings renewed attention to India’s updated Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, which impose stricter due diligence on “significant platforms” Storyboard18.

Under the updated rules, platforms must detect, label, and respond to “synthetically generated information” (SGI), embed metadata identifiers, and implement visible disclosures. Larger platforms are required to obtain user declarations, deploy reasonable technical measures to identify and label synthetic content, and prevent unlawful AI-generated impersonation. Failure to act promptly after receiving a valid court order—within two to three hours for certain categories—can result in loss of safe harbor protection.

Experts caution that the updated rules stop short of a comprehensive legal framework. India lacks a dedicated personality rights statute or standalone synthetic-media offence, leaving courts to define boundaries through individual cases. The Bombay High Court’s decision to allow the suit does not automatically strip safe harbor; private complaints alone do not prove illegality. But the case signals that courts are increasingly willing to examine whether platforms have adopted reasonable technical measures to curb unlawful synthetic media.

Knowns, Unknowns, and Tensions

What is known: Australia has filed a lawsuit against Telegram, the EU’s Coyote System ruling has created a new legal test for algorithmic control, and India’s IT Rules place new obligations on platforms to detect and label synthetic content. What remains uncertain: whether the Australian case will succeed in court; how national courts across the EU will apply the Coyote System ruling; and whether platforms will respond by over-moderation to avoid liability, or by redesigning their recommendation algorithms to maintain immunity.

There are clear disagreements. Telegram insists its anti-terror measures are adequate and will contest the case. The EFF argues that the Coyote System ruling could chill free expression by encouraging platforms to remove lawful content. In India, the tension is between curbing deceptive AI content and protecting political speech and satire. The Bombay High Court suit tests whether safe harbor protections can survive a coordinated, multi-platform deepfake campaign.

The Role of AI Guardrails and the Risk of Over-Censorship

These cases collectively illustrate a growing regulatory push to expand platform liability beyond simple takedown regimes. The EU ruling ties liability to algorithmic design, which could force platforms to redesign recommendation systems or face legal risk. This creates a classic tradeoff: platforms may over-moderate to avoid liability, chilling speech, or under-moderate and face fines. Automated AI moderation systems are a common response, but they introduce their own risks.

A recent incident in India highlights the problem. In July 2026, Meta briefly removed a Facebook post by Prime Minister Narendra Modi that promised stringent action against exam paper leaks. The company attributed the removal to a “technical error” in its AI-driven content moderation system and later apologized. The Indian government demanded an explanation at the highest level, and Meta has since implemented new protocols for high-profile accounts Economic Times. The incident underscores the difficulty of designing AI guardrails that accurately distinguish harmful content from lawful speech, especially when automated filters are applied globally across different political and cultural contexts.

The debate over AI guardrails connects directly to the broader conversation about uncensored versus filtered access to information. Platforms that resist automated moderation—often on free speech grounds—may find themselves in regulators’ crosshairs, as Telegram is learning. Those that embrace aggressive filtering risk false positives and political backlash. The question is not whether to use AI moderation, but how to calibrate it: what thresholds for removal, what appeals processes, and what transparency obligations should apply. The Coyote System ruling, by penalizing algorithmic control, may push platforms to reduce the sophistication of their ranking systems, potentially undermining the very features that make them useful for discovering content.

Implications and Open Questions

These regulatory actions represent a dual challenge to platform immunity. Courts are increasingly willing to hold platforms liable for failing to moderate extremist content, while also treating algorithmic content organization as a form of control that can strip safe harbor protections. The Australia case tests whether fines alone can compel compliance. The EU ruling tests whether algorithmic structure itself can trigger liability. India’s approach tests whether due diligence obligations can be enforced through private litigation.

The open question is whether these approaches will harmonize or create conflicting standards across jurisdictions. For AI-driven moderation systems that are inherently opaque, the risk of regulatory fragmentation is high. A platform that designs its algorithms to comply with the EU’s Coyote System test may find itself struggling to meet India’s detection requirements, or vice versa. The broader implication is that the era of platform immunity may be ending, replaced by a patchwork of liability regimes that demand different forms of algorithmic accountability.

---

FAQ

What is the Australia lawsuit against Telegram about?
Australia’s eSafety commissioner is suing Telegram for allegedly failing to remove pro-terror material, including Christchurch shooting footage, under the Online Safety Act. The regulator seeks a fine of up to A$54.6 million and alleges the platform created a “permissive environment” for extremist content.

What did the EU Court of Justice rule in the Coyote System case?
The EU Court of Justice held that when an intermediary’s algorithm determines “under what conditions, how and in which order of priority” information is disseminated, the intermediary “controls” that information and loses the hosting liability exemption under the e-Commerce Directive.

How does India’s updated IT Rules affect platform liability for AI deepfakes?
India’s updated IT Rules require platforms to detect, label, and respond to synthetic content, including deepfakes, and impose stricter due diligence on significant platforms. The Bombay High Court allowed a civil suit against Meta, X, Google, and YouTube over AI-generated deepfakes, testing safe harbor boundaries.

What is the risk of the Coyote System ruling for free expression?
Critics like the Electronic Frontier Foundation (EFF) warn that the ruling could encourage platforms to proactively monitor and remove lawful content to avoid liability, chilling freedom of expression. The ruling may be difficult to reconcile with the DSA’s acceptance of recommendation algorithms.

How does automated content moderation relate to AI guardrails?
Automated AI moderation systems, like those used by Meta, can produce false positives, as seen in the removal of a post by India’s Prime Minister Modi. This highlights the tension between implementing effective guardrails against extremist content and avoiding over-censorship, a key debate in the design of AI-driven moderation tools.