This is a web version of The Lever’s daily email newsletter.
Thinking of consulting an AI chatbot to research candidates and issues before Election Day? Think again. Political influence campaigns are turning artificial intelligence chatbots into political operatives, exploiting some of the greatest weaknesses of AI models to get their message out to voters. Strategists are encouraging candidates from both parties to publish information that can easily be picked up by AI bots and are creating their own tools to track how chatbots talk to voters about their campaigns. “Imagine knowing which outlets, phrases, or spokespersons are disproportionately influential in the AI’s ‘mind’ — and securing your place in those exact influence nodes,” the D.C. public affairs firm DDC advertises on its website. DDC’s clients include 12 of the Fortune 15 companies and 26 of the nation’s biggest 50 PACs.These campaigns are a part of a larger strategy called generative engine optimization, which involves refining and structuring online information to make it as easy as possible for chatbots to pick up and share. But some actors are adopting more sinister tactics, generating biased election and voter information across various sites disguised as neutral news sources. This so-called pink slime content isn’t necessarily intended for human consumption. Instead, AI chatbots cite it when prompted for information about candidates. It has often come with a boomerang effect, where real information is dismissed as AI.
An Israeli intelligence firm recently found that someone set up 37 fake local news sites around the country on a single night in September. With names like Wolverine State Report and Electric City Watch, the sites were filled with oddly written information and hidden code designed solely to be discovered and read by AI chatbots — and the sites all agree with each other on which local candidates are best.(VisionHeight)“Local news has been shrinking for years,” wrote the researchers. “In that vacuum, four outlets that agree look like a consensus.”Many of these sites are now highly rated on Google, above Ballotpedia. (VisionHeight)Pink slime has its similarities to misinformation campaigns political strategists have used in the past. One similar, analog approach in California published dossiers, sometimes containing false information, about Democrats to conservative partisan news outlets run by Republican operatives.But as AI takes over, the spin machine has shifted into overdrive. Roughly one in five voters report using chatbots to learn about elections. And a recent audit published by Politico found that leading AI chatbots are citing pink slime content in nearly half of all answers to midterms-related queries.Talk about slimy.
DAY IN DATA
The GOP is the party of the rich and politically active. (Source: CNBC)
RONNIE’S REPORT
🤫 Anthropic says the quiet part out loudThe artificial intelligence giant Anthropic has released its IPO paperwork to go public on the stock market, and the numbers are sobering. After losing $42 billion in 2025, the maker of the chatbot Claude reported $80 billion in net losses in the first half of 2026. During that time, just under 50 percent of Anthropic’s revenue came from contracts with Amazon, Google, and Microsoft — firms that are also major investors.Anthropic also has more than $500 billion in planned infrastructure and cloud expenditures in the coming years. Still, the company is targeting a historic IPO valuation of as high as $2 trillion, potentially topping SpaceX’s recent record-breaking IPO.It’s Not An AI Bubble. It’s A Black Hole.How $400 billion in AI spending became a gravitational force that’s distorting the entire economy.While reiterating its commitments to AI safety and responsibility, Anthropic also laid out some startling risk factors inherent to its business. “Models, including our own, have exhibited, and may in the future exhibit, unexpected strategic or self-preserving behaviors, including attempts to resist shutdown, conceal or manipulate information, or generate outputs that could be interpreted as coercive, deceptive, or manipulative,” the filing says, “including behavior resembling blackmail or efforts to influence human decision-makers to preserve the model’s continued operation or access.”The fallout could be uninsurable, says Anthropic: “With the scale of harm that could result from a catastrophic AI-related incident, no single insurer, and potentially the insurance industry in the aggregate, may have sufficient capacity or expertise to underwrite insurance commensurate with such exposure.”But in a recent interview with The Lever, industry expert and AI critic Ed Zitron poured water on AI doomsday scenarios. “It’s not dangerous because AI has gone rogue. It’s dangerous because we have OpenAI, Anthropic, Meta, and multiple other companies that are running these dangerous experiments connected to hundreds of billions of dollars of AI infrastructure provided to them by the largest companies in the world,” said Zitron. Cutting Through the AI Doomsday NoiseAn exclusive conversation in which David Sirota talks with Matt Stoller and Big Tech critic Ed Zitron about whether we’re staring down an AI apocalypse.Anthropic somewhat seems to agree. “AI deployed at scale also has the potential to concentrate power and wealth both within and between countries in ways that could erode trust, harm society, and destabilize the geopolitical order,” the IPO paperwork admits.
Got scam calls? The worst part hasn’t happened yet.When scammers get hold of your data, they don’t just try to take your money. They can sell what’s left — your name, address, phone number, the names of people close to you — to the next buyer.Earlier this year, Congress put a number on it: four breaches alone have cost Americans $20 billion.Incogni removes your data from those databases, protecting your finances and your personal information. Get 55% off with code LEVER.
YOU LOVE TO SEE IT
💬 Everybody loves the sunshineA judge ruled that Texas Gov. Greg Abbot (R) must turn his communications with Elon Musk over to his Democratic gubernatorial opponent, state Rep. Gina Hinojosa. The judge agreed that Hinojosa has a right as a member of the Texas House of Representatives to inspect the communications, which would have otherwise been shielded from public review, for legislative purposes. Abbott has appealed the order that he hand over more than 1,000 unredacted pages of his personal communications with Musk. It’s unlikely that the paperwork will be shared by the public before Election Day.
Create a FREE account, or log in.
Registered readers of The Lever get access to a limited number of free articles every month, breaking news alerts, and our Lever Daily newsletter.
Already have an account? Sign in
Click Here For The Original Source.
——————————————————–
