Edition 13 · September 18, 2026
Anthropic published a report documenting how state actors and criminals misused its Claude AI. Mali’s junta used it to help run a surveillance system tracking 25 million SIM cards. Russian, Iranian, and Yemeni groups used it for drone code, propaganda, and weapons design. Anthropic says it disrupted the accounts, and its CEO called for the industry to slow down. Separately, a German court ruled Meta liable for fraudulent ads on Facebook and Instagram.
The Anthropic Report
- Anthropic’s report details how Claude was used to scale fake newsrooms, surveillance, and scams (Business Standard, Sep 11), spanning roughly 70 fake news sites, each with a fake byline and X account, boosted by hundreds of comment accounts.
- Mali’s military junta used Claude to help build “Lakana 360,” a system monitoring 25 million SIM cards (Japan Today, Sep 13); Anthropic says it could not disable the system once deployed.
- A Russian group tied to the SVR intelligence service used Claude to write espionage tools and code for autonomous drones (Inquirer, Sep 11) aimed at Ukraine.
- Iran-linked groups used Claude to target US Navy forces and run propaganda (Times of India, Sep 13), and Anthropic separately disrupted Iranian use of Claude to spy on dissidents (RFE/RL, Sep 11).
- A UAE-linked operation used Claude to run ~300 fake accounts and a front NGO (Middle East Eye, Sep 11) shaping discourse on the Muslim Brotherhood and the Sudan war.
- One person ran a pro-Awami League fake news network in Bangladesh for nearly a year (The Print, Sep 11), generating 1,500+ headlines with Claude.
- Anthropic CEO Dario Amodei “We must slow the pace” (Moneycontrol, Sep 12) after the disclosures; China rejected Anthropic’s related claim (NewsBytes, Sep 12) that its labs distilled Claude.
Election Integrity
- Anthropic said it disrupted an AI-powered election manipulation platform targeting Malaysia (Independent, Sep 11) that used 1,000+ fake X accounts across all 222 constituencies.
- German intelligence reports a rise in Russia-linked disinformation targeting Chancellor Friedrich Merz (Yahoo, Sep 16) since June.
- Analysts identified 8,500+ suspicious X accounts trailing Swedish political figures (Euronews, Sep 16), many claiming to be based in West Africa and Asia.
- In Kenya, an AI operation manufactured online support for a cabinet secretary (Nation, Sep 11) ahead of the 2027 election.
State Propaganda & Info Warfare
- The Institute for the Study of War found at least 35 suspected AI-faked videos of Russian territorial gains in Ukraine (Kyiv Post, Sep 15), with manipulation confirmed in 10.
- A Newsweek investigation traced China-made social posts that sparked anti-US-base protests in America (Newsweek, Sep 16), tied to a network funded by Neville Roy Singham.
- The British royal family is a target of fake news from foreign “state-funded” bot farms (The Times, Sep 16).
- China’s state security minister warned of “cognitive warfare” using generative AI (Times of India, Sep 14).
Platform Accountability
- A German court ruled Meta liable for fraudulent third-party ads (Times Now, Sep 17), ordering removals and damages after a brand was used in fake investment posts.
- Meta’s Oversight Board called the company’s deepfake rules “consistently and fundamentally inadequate,” (Engadget, Sep 17) and separately urged Meta not to replace fact-checking with Community Notes (The Standard, Sep 12).
- Brazil’s attorney general notified YouTube to remove AI videos impersonating doctors (Digital Watch, Sep 11), citing 97 profiles pushing false health claims.
Fact-Check Frontlines
- Fabricated newspaper front pages are spreading fake campaign promises in Nigeria (Yahoo, Sep 15) ahead of the January 2027 election.
- A viral video of an “American AI robot escaping a lab” was debunked as fully AI-generated (Latestly, Sep 14).
Scams & Fraud
- Fintech Revolut disclosed a breach after handing customer data to actors impersonating government agencies (BleepingComputer, Sep 14), exposing passports and financial records.
- TikTok users are being sold fake GLP-1 weight-loss patches (Boston Herald, Sep 15), often receiving a different product than advertised.
- AI-generated “wellness influencers” are spreading dubious health claims (Marie Claire, Sep 12), earning far more than human creators while pushing unverified advice.
Quick Hits
- A Brazilian YouTuber is using an AI-generated Australian voice to promote the One Nation party (ABC, Sep 16) to 3M+ viewers; the party denies any link.
- In India, police arrested a man for deepfake fraud using videos of Tamil Nadu CM Vijay (Times of India, Sep 16) and warned that the viral “1980 avatar” selfie trend risks biometric harvesting (Times of India, Sep 12).
- Hong Kong announced plans to legislate against AI-enabled crime, including fraudulent deepfakes (The Standard, Sep 16).
Try This — Bad Aura
- A recruiter reading your feed isn’t looking for your best posts. They’re looking for reasons to pass. A scammer isn’t either: they’re looking for an opening.
- Bad Aura, from Aunoo, connects to your account and reads your public posts back to you through their eyes: what a recruiter flags for (un)employability, and what a scammer sees to work with.
- Same posts, two verdicts you never get to see. Now you can.
- See how they see you →
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
AI Transparency
This newsletter is produced with AI. Aunoo uses automated systems to gather, score, and draft the content, with human review before publication. In line with the EU AI Act’s transparency obligations (Article 50), which took effect August 2, 2026, we disclose that parts of this edition were generated or assisted by artificial intelligence. Every claim links to its original source — please verify against those before reuse.
