The fight this week is over who controls online speech, governments or platforms. India blocked all of Telegram for 150 million people instead of removing individual posts it objected to. Leaked “Project 2026” files show Russia building its own network of fake reference sites and friendly outlets, while China reportedly used ChatGPT to seed influence campaigns. The sense is that authoritarian states keep getting better at manufacturing information, and democracies keep getting clumsier about policing the platforms that carry it. This weeks deep dive is on India’s ban. Also inside, how much of TikTok’s feed is now AI sludge, and a US official’s reported pitch to annex Greenland for shrimp.

Platform warfare

India didn’t pull posts off Telegram. It pulled Telegram. The Delhi High Court upheld the government’s order to block the entire app under Section 69A of the Information Technology Act, cutting off 150 million users. The stated justification was NEET exam fraud, though the government’s filing extended further, citing child abuse alongside scams and pirated content.

The ban was scheduled to lift on June 22, but users continue to report access problems, indicating either an incomplete technical restore or continued restriction. Digital rights groups warn the order expands platform-level censorship.

What to watch: when a democracy this large can turn off a whole app, smaller governments get the cover to try the same thing.

Disinformation & state propaganda

• Leaked documents indicate Russia’s Social Design Agency (SDA) is running “Project 2026,” a plan to build a parallel information network of Wikipedia-style reference sites and aligned media outlets, supported by fake think tanks. The SDA, sanctioned in the US, UK and EU, targets both public belief and the training data that feeds AI chatbots.
• Russia also continues to work against Ukraine’s EU accession through online disinformation campaigns exploiting concerns over corruption, security and economic cost. A joint Ukraine-EU report documented 80 incidents between January and May 2025, each tailored to a specific national audience.
• Analysts also flag potential Chinese Communist Party influence in the 2028 Philippine presidential election, weighted toward digital campaigns and local operations rather than overt interference.

AI’s credibility crisis

• AI-generated video now dominates parts of TikTok’s feed. A Kapwing study found 294 of 500 For You videos were AI-made, about three times YouTube’s rate. Exposure is highest for minors: 57% of children’s content was AI-generated, and new accounts saw 59%.
• Cornell Tech researchers identified a vulnerability in AI research tools: a single Reddit comment can steer ChatGPT and Gemini toward fraudulent services. Between 54% and 71% of the content influencing these systems originates on Reddit, creating a new attack surface for fraud.
• Meta’s Oversight Board is pressing for stronger protection against sexualized deepfakes of non-public figures, urging the company to add AI-generated impersonations to its Adult Sexual Exploitation policy.

Scams & digital crime

• An MIT investigation documents Telegram operating as a marketplace for AI fraud tools, including software marketed to defeat facial recognition and Know-Your-Customer checks. The platform’s group structure and encryption lower the operating cost for organized crime.
• The Department of Justice announced the seizure of cloud infrastructure linked to the Huione Group, a Southeast Asian conglomerate previously cut off from the US financial system.
• Polymarket drew scrutiny after investigators found its viral “big win” videos were staged on cloned websites. The highlighted bets lost money.

Misinformation in motion

• Nigerian police rejected viral claims that abducted schoolchildren and teachers had been flown to Kaduna by helicopter, calling the video “false and misleading” and confirming that rescue operations continue.
• The DOJ’s release of 3.5 million pages under the Epstein Files Transparency Act has driven new conspiracy theories. Phased releases of heavily redacted documents have sharpened questions about completeness.
• Research quantifies the asymmetry: sunscreen misinformation outperforms accurate posts on social media, with claims that sunscreen is harmful drawing more likes, comments and shares on TikTok.

The surveillance question

• Critics argue age-verification systems function as mass surveillance rather than child protection, because they require identity checks on every user. The UK government is weighing VPN restrictions within its broader “age-gate” proposals.
• The Electronic Frontier Foundation backs new legislation targeting government pressure that leans on platforms to suppress lawful speech, a form of indirect censorship conducted through regulatory threat.
• Export controls on cyber tools face similar criticism. Analysts argue that 30 years of evidence shows the restrictions do not work, with Anthropic’s Mythos cybersecurity model the current test case.

Media & journalism under pressure

• The UK’s Labour government is considering rules that would require newspapers to join state regulators, paired with social-media controls obliging platforms to surface public-service broadcaster content. The proposal reportedly covers national and local papers.
• Newsguard faces criticism for rating Chinese state media as more reliable than American conservative outlets while building AI tools to counter misinformation and selling credibility scores to advertisers for ad placement.
• ABC is asking viewers to oppose an FCC effort to control who appears on The View, characterizing the oversight as content control.

Conspiracy theories & declassified chaos

• A federal judge rejected Biden’s request to block release of recordings made with a ghostwriter, ruling that public interest outweighed his privacy claim in the Heritage Foundation’s request.
• New polling shows bipartisan demand for additional UAP disclosure amid low trust in federal institutions, while declassified files allegedly link Russia and China to reverse-engineering of crashed technology, citing incidents including the Harare event.
• Trump endorsed comparisons to Hitler, Mao and Stalin, telling a golf partner the assessment “sounds good to me” and adding that the “overwhelming difference” is that he is more powerful.

Power plays & political theater

• The administration’s handling of Anthropic’s Mythos AI model continues to escalate. Critics contend the government is pressuring the company for declining certain demands, with potential cost to the cybersecurity community.
• Financier Stephen Feinberg has accumulated significant influence over US military spending from inside Trump’s Pentagon, operating largely out of public view.
• Federal Reserve governor Lisa Cook incurred more than $1.3 million in legal and security fees after the administration moved against her. The figure appeared in ethics disclosures while she remains central to a Supreme Court case over the legality of her firing.

Quick hits & oddities

• A Harvard study measured 38 years of misinformation coverage in major US dailies, tracing how the press established misinformation as a standing beat.
• Google’s YouTube settled a social-media addiction case brought by a 15-year-old; suits against three other tech companies proceed next month.
• A Trump administration official reportedly proposed annexing Greenland to secure shrimp supply for Red Lobster, citing the island’s seafood resources and reduced reliance on China.

---

The deep dive: Platform warfare

India’s Telegram ban signals a new era of platform-level censorship

India’s June 2026 Telegram ban is the first time a democracy this large has switched off an entire encrypted app instead of going after the accounts or posts behind the problem. The Delhi High Court upheld the nationwide block under Section 69A of the Information Technology Act, cutting off 150 million users and stretching what counts as a normal government response to platform misuse.

The development. India blocked Telegram nationwide from June 21 to 30, 2026, officially to prevent cheating on the NEET-UG medical entrance retake. The government’s justification extended past exam fraud to child abuse, financial scams and pirated content.

Consensus and contradictions. Sources agree the Delhi High Court upheld the ban under Section 69A and that it sets a precedent for blocking platforms whose “architecture hinders targeted enforcement.” Where they diverge is on the timeline: the ban officially ended June 22, yet users still report problems. The sharper division is over framing. Officials stress criminal misuse; digital rights observers warn of expanding platform-level censorship that could extend to Signal, WhatsApp and other encrypted services.

Broader implications. Other governments writing rules for encrypted messaging now have a working model to point to. Once a court accepts that a whole app can go dark because some of its users break the law, that logic travels well. And given India’s size, a ruling over 150 million users won’t stay inside India’s borders.

Timeline

• May 21, 2026 — India’s National Testing Agency flags Telegram to MeitY for circulating leaked NEET-UG exam papers. (SCC Online)
• Jun 16, 2026 — MeitY issues a platform-wide block order under Section 69A — effective within an hour, running to Jun 22; ~150 million users cut off. (Social Samosa)
• Jun 19, 2026 — The Delhi High Court (Justice Tejas Karia) upholds the block, ruling that a whole app counts as “information.” (Bar & Bench)
• Jun 22, 2026 — The platform block lifts (message-edit feature stays disabled until Jun 30), but users keep reporting access problems. (Internet Freedom Foundation)

What to watch:

• Legal challenges to other encrypted platforms: whether authorities apply the precedent to Signal, WhatsApp or others on similar grounds.
• International platform responses: how Telegram and competitors adjust their India operations and verification requirements.
• Restoration completeness: whether Telegram service fully returns or remains degraded, an indicator of continued pressure.

---

About Bad Information

A weekly read on disinformation, coordinated influence operations, and media literacy. Researched and fact-checked with Aunoo.

Powered by Aunoo – news intelligence with trust built in. Aunoo scores every article on five trust signals before you even click: source bias, factual-reporting rating, AI-content detection, disinformation-network flags, and a quality score. You see how far to trust a story, not just what it claims.

Try it, it’s free → saas.aunoo.ai