Edition 6 · July 31, 2026
Researchers caught Google Gemini citing a fake NGO founded by a former Wagner Group boss as if it were a real source. That’s the shift worth watching this week: the target isn’t only voters, it’s the AI people increasingly use to check facts. Russia is quietly poisoning the data these chatbots pull from, so they repeat Kremlin narratives as fact. Around it, rival governments trade deepfake accusations across borders, the EU’s deepfake-labeling rules take effect Sunday, and an Iran-linked video openly threatened Melania Trump. The tools built to help us sort truth from noise have become a target in their own right. Inside: Russia’s AI poisoning, Telegram’s legal wars, and a fresh look at who’s really behind this week’s “foreign interference” claims.
Deepfake Diplomacy
- In China, AI-generated disaster videos (BBC) are spreading panic and complicating rescues; authorities have started arresting the creators.
- The Indian government alleges it found 500+ social-media profiles spreading AI-generated content around recent protests (Times Now), attributing about 100 accounts to actors outside India (The Hindu), mainly Pakistan and the Middle East. The attribution is single-source and unverified, and Pakistan makes a mirror claim about India (see Consensus Check, below).
Consensus Check: The Maharashtra “Foreign Disinformation” Claim
- We ran the lead item through Aunoo’s independent validation. Verdict: contested (0.67).
- The allegation comes from one source, the Indian government, and spread as near-identical copies across outlets rather than as independent reporting. No published forensic evidence supports the Pakistan/foreign attribution.
- The outlets don’t even agree on the event: The Hindu calls it the NEET protests; News18 says the network targeted “Cockroach Janata Party” protests.
- Pakistan alleges the reverse, an India-linked network funding campaigns in Azad Kashmir. Both governments accuse each other, and neither claim is independently verified.
- Bottom line: a real inauthentic AI network, yes; confirmed foreign interference, no. It reads as a government allegation, not a validated finding.
Election Interference
- A former Surrey mayor wants a federal probe into alleged foreign interference in BC municipal elections (CTV) via India-linked Facebook pages.
- Ex-US officials question whether Trump shaped a CIA assessment (Firstpost) to back China-interference claims, even as his China rhetoric softens around TikTok (Fox News).
- Israel’s Shin Bet chief warned of Iranian attempts to meddle in Israeli elections (Times of Israel), while reporting continues on Israel’s own $50M AI-driven US influence campaign (Jerusalem Post).
Propaganda Poisoning
- Researchers say Russia is “poisoning” AI models via RAG poisoning (Head Topics), editing the sources chatbots retrieve so they repeat false narratives.
- Google Gemini and other LLMs were caught citing an EU-sanctioned pro-Russian outlet (Euronews) — a fake NGO founded by a former Wagner leader.
- An investigation exposed a fake Singapore think tank (IEASS) used for espionage (Moneycontrol), and France ordered Russian commentator Xenia Fedorova to leave (Firstpost) over Kremlin propaganda.
Platform Accountability
- European researchers accuse TikTok, X, and Meta of stonewalling data access (Ars Technica) required under the EU’s Digital Services Act.
- California Democrats reintroduced an AI Ads Act (NY Post) to ban AI impersonation of candidates, after viral AI attack ads from figures on the right.
- The EU AI Act’s deepfake-labeling rules take effect Sunday, August 3 (Yahoo), with large fines for non-compliance.
Regulatory Reckoning
- India approved 13 deepfake-detection projects and cut the takedown deadline to three hours (Times Now).
- TikTok published election rules for Israel’s Knesset vote (Jerusalem Post), banning paid political ads and unlabeled AI content.
- Researchers proposed AISPA (arXiv), a way to audit the hidden system prompts that govern how AI apps behave.
Scams & Impersonation
- A viral Rajasthan “hotel” video was staged clickbait (Stack Umbrella) built to funnel viewers into phishing links and Telegram channels.
- The FakeAgent campaign used Bing ads and a malicious Claude Artifact (OnMSFT) to push fake Claude installers carrying a remote-access trojan, hitting 29+ organizations.
- Security teams are now running deepfake phishing simulations (TechTarget) that clone executives’ voices and faces to test staff.
The Telegram Wars
- Russia’s FSB charged Telegram’s Pavel Durov with aiding terrorism (The Star) and added him to wanted lists; he replied with a middle-finger post.
- Ukraine says it detained 46 Russians aged 12–22 (TASS) recruited via a Telegram bot into infrastructure attacks, and Australia’s eSafety Commissioner is suing Telegram over terror content (SBS).
Health Misinformation
- Canada lost its “measles-free” status (The Conversation) as vaccination rates fall amid rising vaccine disinformation.
- Researchers note vaccine skeptics out-storytell public-health messaging (NZ City), and BBC/AFP debunked the claim that sunscreen causes cancer (BBC/AFP).
Fact-Check Frontlines
- South Korea’s heavy rains revived debunked claims that China is “cloud-seeding” the weather from the Yellow Sea (Yahoo).
- Malaysia’s Election Commission debunked a fake early-vote infographic (The Star), and Delhi Police flagged 400+ Pakistani accounts spreading NEET-protest misinformation (Asianet).
Quick Hits
- Antisemitic content is nearly 7x more likely to top Instagram clip accounts’ most-viewed videos (Jerusalem Post), per an Antisemitism Research Center study.
- The House Ways and Means Committee subpoenaed parts of the Singham network (NY Post) over CCP propaganda and anti-data-center efforts.
- A cryptographic fix, STEBR (Tech Times), addresses the Signal backup-key flaw Russian spies exploited.
- An Indian minister is suing Meta, X, and Google over AI deepfakes (Deccan Herald) that tied him to corruption, an early test of whether defamation law can handle synthetic content at scale.
Notable Mentions
- A “Where to Kill Melania?” video on Iran-linked Tasnim News (Moneycontrol) allegedly threatened the US First Lady and Barron Trump.
- A federal judge said the Trump administration lacks evidence to label Anthropic a “supply-chain risk,” (TechCrunch) undercutting its ban.
- New multilingual AI detects misinformation in isiZulu and Sepedi (iAfrica), narrowing a language gap in fact-checking tools.
Curious how a recruiter may view your social media activity?
A recruiter reading your feed isn’t looking for your best posts. They’re looking for reasons to say “Pass!”. A scammer isn’t either. They’re looking for an opening.
Bad Aura reads your public Bluesky posts back to you through their eyes, showing you the sort of things AI used by recruiters flags for (un)employability, and what a scammers see to work with. Entirely free, as in “Free Beer” 🍺
See how they see you → Badaura.aunoo.ai
About Bad Information
A weekly read on disinformation, coordinated influence operations, and media literacy. Researched and fact-checked with AuNoo AI .
Powered by AuNoo AI , the open strategic intelligence platform 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
AI Transparency
This newsletter is produced with AI. Aunoo uses automated systems to gather, score, and draft the content, with human review before publication. Parts of this edition were generated or assisted by artificial intelligence. Every claim links to its original source. Please verify against those before reuse.
