Two stories dominate this week. Meta released Muse Image, a tool that generates AI images from any public Instagram photo by default. Users must opt out to keep their likeness from being manipulated, and nobody is notified when it happens. And President Trump fired the remaining members of the bipartisan Election Assistance Commission, the body that maintains the federal mail-voter registration form, months before the midterms. This edition covers both, plus OpenAI accused of hiding evidence in its copyright trial, marketing agencies seeding Reddit with 25,000 poisoned posts a day to manipulate AI training data, fake IT staff installing malware over Microsoft Teams, and Christian missionaries running prayer circles in VRChat.

Election Interference & Voting Rights

• Trump fired the remaining members of the bipartisan Election Assistance Commission (NPR, July 9), which maintains critical election administration functions including the federal mail-voter registration form. Democrats and voting rights advocates condemned the move, which comes months before the midterms.
• The administration is threatening states with criminal charges in fights over voter data (The Guardian, July 8), demanding information about noncitizen voting despite unproven claims. State officials called the federal demands “truly bizarre behavior.”
• New voting barriers are emerging across multiple states (NPR, July 8), with journalist Ari Berman warning that many Americans do not have easy access to required documents like birth certificates. Redistricting, campaign finance changes, and new voting requirements are rewriting the rules for the midterms.
• Spain arrested an alleged supporter of pro-Russian hacktivist groups (The Record) following an FBI tip linking the individual to CyberArmy of Russia Reborn, Z-Pentest, and NoName057(16).

AI-Generated Manipulation

• Meta’s new Muse Image tool can generate AI images from any public Instagram photo (CNET) by default. Users must opt out rather than consent, and get no notification when their likeness is used; privacy campaigners called it “a recipe for disaster” (BBC), and India’s IT secretary said the government will examine the tool (Economic Times) for legal compliance.
• Google’s deepfake detector debunked a fabricated image of Senator Mitch McConnell (TechCrunch, July 8) that showed him distressed in a hospital bed. The fake targets a politician whose real, unexplained four-week hospitalization is already feeding speculation.
• A class action against SpaceXAI and Stability AI has expanded (NPR, July 9), with new plaintiffs alleging the companies’ tools were used to create sexually explicit images of them as children.

Disinformation & Fact-Checking

• Harvard’s Misinformation Review published research on narrative monitoring and argument-checking (Misinformation Review), using Spain’s October 2024 floods as a case study where false claims about concealed casualties gained traction despite video evidence of evacuations.
• A harms-based model for choosing which claims to fact-check (Misinformation Review) has been tested across Europe, Africa, and the Middle East since 2024, ranking claims by the imminent and cumulative harm they could cause.
• European fact-checkers are setting new standards for correcting their own errors (Misinformation Review), going beyond the compliance requirements of the International Fact-Checking Network and the European Fact-Checking Standards Network.
• New research demonstrates detecting information operations through behavioral and language patterns (arXiv, July), using unsupervised anomaly detection to spot coordinated inauthentic behavior without relying on oversimplified assumptions about how operators coordinate.

Platform Accountability & Content Moderation

• X will send users a direct message when a post they engaged with gets a Community Notes correction (TechCrunch, July 8), addressing the criticism that crowdsourced fact-checks usually arrive after the misinformation has already spread.
• Reddit’s AI systems are flagging 25,000 poisoned posts a day (Tech Times, July 8), seeded by marketing agencies to influence what ChatGPT and Gemini say about their clients. Because AI systems lean on Reddit for training data, the site has become an attack vector for manipulating model outputs. The 25,000 figure is single-sourced to Tech Times.
• The EU Parliament approved “Chat Control 1.0” (patrick-breyer.de), extending permission for companies to scan citizens’ private messages for child abuse material.
• A Bondi Beach terror attack victim criticized YouTube (ABC News Australia, July 8) for keeping up a video that falsely accuses him of lying about his injuries.
• Google will now disclose which ads are made with AI (TechCrunch, July 9), labeling ads that advertisers created or edited with generative AI tools.

Press Freedom & Legal Battles

• A federal judge threw out Trump’s $3.8 billion defamation lawsuit against the Washington Post (The Guardian, July 8) over a 2023 article about a bank with pornography-industry ties helping fund Trump’s media operation.
• News outlets urged a judge to sanction OpenAI for allegedly hiding evidence (AP News) in the copyright fight over ChatGPT. The New York Times claims OpenAI faked its inability to search training data and deleted billions of logs (TechCrunch, July 9).
• A US appeals court struck down key parts of Florida’s Stop Woke Act (The Guardian, July 7), another legal setback for the law restricting campus discussions of race and gender.
• ABC is fighting the FCC over “The View” and equal-time rules (AP News), challenging the regulator’s interpretation of broadcast fairness requirements.
• South Korea’s law targeting “fake news” took effect (AP News) as journalists’ groups raised press-freedom concerns about how it will be enforced.

Surveillance & Privacy

• Madison Square Garden kept a database tracking and categorizing celebrities (WIRED), with labels including “LGBTQIA,” “DO NOT HOST,” and risk levels, covering hundreds of celebrities, Knicks superfans, and Taylor Swift’s wedding guests.
• Suspected Chinese operatives broke into universities’ Roundcube mail servers (The Register, July 8) in campaigns affecting dozens of targets, according to Proofpoint researchers.
• A lawsuit alleges the US illegally shared confidential information on Iranian asylum seekers with Iran (AP News), exposing the people involved to retaliation.
• Hyderabad police deployed a social media monitoring platform called SOCEYE (SecurityBrief) that tracks online activity in multiple Indian languages and preserves digital evidence for investigations.

Cybercrime & Social Engineering

• Fake IT staff on Microsoft Teams are coaxing workers into installing malware (The Register, July 7), posing as helpdesk personnel to gain remote control before dropping the EtherRAT trojan.
• A threat actor dubbed Lurking Lizard has run a malicious residential-proxy business since 2022 (The Hacker News), using fake 7-Zip installers across more than 230 lookalike domains to turn victims’ devices into proxy nodes.
• North Korea-linked npm packages are mimicking Rollup polyfills (The Hacker News) to steal developer secrets, copying legitimate project metadata to look authentic.
• A Reddit scam sends users fake “your account was reported” messages (Help Net Security, July 9), then extracts login credentials through social-engineering conversations backed by fabricated evidence.
• Wellness influencers are promoting $50 “EMF straws” (WIRED) claimed to block electromagnetic radiation from electronic devices. There is no scientific evidence they do anything.

Political Scandals & Corruption

• The former CFO of Epoch Times pleaded guilty in a $67 million money laundering scheme (AP News), a significant financial-crime case involving the media organization.
• The Guardian’s investigations into Reform UK’s crypto donations keep getting pre-empted by friendly stories (The Guardian, July 9), and Nigel Farage faces questions over donations from a convicted fraudster (AP News).
• Erika Kirk asked a judge to make all exhibits in her husband’s killing case public (The Guardian, July 9), explicitly to head off conspiracy theories.
• Questions continue over Mitch McConnell’s hospital stay (BBC), now past four weeks with no disclosed reason. That vacuum is what the fabricated hospital-bed image above stepped into.

Extremism & Violent Threats

• Hundreds of masked Patriot Front members marched through Washington DC on July 4th (The Guardian, July 4), carrying Confederate flags through the capital on the nation’s 250th anniversary.
• Eight men were indicted on murder and terrorism conspiracy charges (AP News) for their alleged roles in a thwarted drone and sniper attack on Trump’s UFC event at the White House in June.
• Germany’s AfD is positioning itself for regional election gains (AP News), buoyant despite protests and eyeing a share of power in upcoming votes.

Misinformation & Conspiracy Theories

• The leaked Epstein files are feeding viral rumors and conspiracy theories (Film Daily). Roughly half of the estimated six million pages remain withheld or heavily redacted, which keeps the speculation supplied.
• The Smithsonian’s secretary emphasized “accuracy and integrity” (NPR, July 9) after a White House report accused the National Museum of American History of “radical, activist ideology.”
• An Idaho mother who claimed her twins died from vaccines was charged with their murder (The Guardian, July 6). Andrea Shaw and her husband had appeared on an RFK Jr-linked podcast after the 18-month-olds died in May last year.

The Weird & Wonderful

• The French national soccer team flew to World Cup matches on ICE deportation planes (The Guardian, July 9). The aircraft have flown 44 deportation flights this year, and the squad’s star, Kylian Mbappé, is outspoken against the far right.
• Christian missionaries are evangelizing in VRChat (AP News), with the evangelical organization Cru running virtual prayer circles where stormtrooper and orange-cat avatars lead religious discussions.
• More than 400 people were arrested at a Fourth of July “TikTok takeover” (The Guardian, July 7) after thousands descended on Newport Beach for a flash-mob meetup that sparked fires and required mounted police.
• An AI cheating scandal is roiling Brown University (Ars Technica), where one professor argued that “we cannot choose to become idiots.”

Notable Mentions

• British officials say technology is driving an increase in online threats to the UK (AP News), particularly from Russia and Iran.
• A former DeepMind executive warns the AI arms race could end in disaster (WIRED), citing the US government’s nationalistic approach.
• The Southern Poverty Law Center pleaded not guilty to federal fraud charges (AP News) over allegations it defrauded donors by not disclosing payments to informants inside extremist organizations.
• Truecaller clashed with India’s telecom regulator over anti-spam rules (TechCrunch, July 8).

---

Situation Report Panzer Parrot: AfD’s “Alternita”, a Party-Run AI System for Mass-Producing Rage Bait

• Germany’s AfD has built an in-house AI system called Alternita that turns far-right news articles into ready-to-post social content in minutes. An undercover Correctiv journalist, given a demonstration and a test account, watched it generate posts demanding forced deportations (“remigration”) and, from a neo-Nazi magazine article, a post warning of an LGBT “rainbow dictatorship” with an AI-generated image of a burning rainbow flag.
• The tool runs on Google Gemini, OpenAI’s ChatGPT, and Anthropic’s Claude. Correctiv found all three engines in Alternita’s source code. Asked whether this complies with their terms of service, Google and OpenAI said they are examining the use. Anthropic did not reply.
• The platform is registered to AfD general secretary Hans-Holger Malcomeß and, per Correctiv, is already running on co-leader Alice Weidel’s channels, where not all output is labeled as AI-generated. The pilot launched the week of July 6, full rollout is planned for 2027, and a September state election where the AfD polls around 40% comes first.
• What distinguishes this from a covert bot operation is that a major party is running centralized propaganda production on mainstream commercial AI services, with the output distributed through party accounts and a base that looks organic. It is also a live test of whether AI providers’ usage policies against political manipulation can be enforced in practice; so far the exposure came from journalists, not from the companies’ own monitoring. Aunoo tracks the campaign as PANZER PARROT.
• Confidence: high on the tool’s existence, capabilities, and AI engines, all documented through Correctiv’s demonstration and source-code analysis; moderate on the current scale of live deployment, which rests on Correctiv’s assessment against the party’s “pilot” framing.
• Read the full situation report → · Source SITREP on GitHub

---

Stay skeptical. Especially of the confident stuff.

---

About Bad Information

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

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