machine-readable.media

Topic

News

AI Search

Grok and GPT’s opposing election picks expose AI’s source-selection power

The same voting prompt produced sharply different recommendations—and sharply different information diets.

A study that prompted the models’ APIs across nearly 500 US midterm races found Grok 4.7 chose Republicans in 92% of answers, while GPT-6.1 Sol chose Democrats in 97%, with agreement below 10%. Their consumer apps refused direct voting advice, a distinction OpenAI emphasized, but the models also searched different issues and source mixes before deciding. As chatbots become election-information interfaces, those opaque retrieval choices can redistribute visibility and influence among news, reference and social sources.

AI Search

AI election answers are influencing donations and creating a campaign-optimization market

Search summaries and chatbots are becoming election intermediaries—and campaigns are learning to optimize for them.

Nearly 40 percent of survey respondents had recently received election information from Google’s AI-search summaries, while 4 percent had consulted a chatbot. The report traces three users’ $1,011 in misdirected donations to a misleading page elevated by Google’s AI, as campaign consultants begin monitoring model answers and publishers’ crawler blocks and licensing deals narrow the source pool. As campaigns optimize for answer engines, those interfaces gain more control over which information sources reach voters and where political spending flows.

News

One in five voters consult AI chatbots for election news, shifting discovery beyond local media

Chatbots are filling gaps in down-ballot coverage while keeping their source choices largely out of view.

A September survey cited by NPR found that about one in five voters consult AI chatbots for election news and voting, with similar use among Democrats and Republicans. Voters told NPR they use OpenAI’s ChatGPT, Google’s Gemini and Anthropic’s Claude to research thinly covered local races, compare candidates and check claims, even as researchers warned that source selection is opaque and agreeable answers can reinforce prior views. As voters delegate political research to chatbots, local publishers risk losing both first contact with audiences and influence over which reporting enters the decision process.

News

Philadelphia Inquirer makes AI lead-finding critical infrastructure for local newsletters

An editor-trained monitoring system now feeds six local newsletters and is being expanded across the newsroom.

The Philadelphia Inquirer’s Scrape runs AI over editor-curated municipal, school and social sources to produce hyperlocal tip sheets, after one editor had spent about 12 hours a week finding newsletter items. Following three to four months of annotated feedback, the tool is now “load-bearing and critical infrastructure” for six newsletters, soon eight, with nearly 20 journalists following its output. The rollout shows how a publisher can turn editorial judgement into discovery infrastructure that expands local coverage capacity without asking an already stretched desk to absorb more monitoring work.

News

Chatbots cite partisan local-news lookalikes in nearly half of tested answers

The study shows how source selection inside answer engines can turn low-transparency political outlets into apparently authoritative news suppliers.

A NewsGuard audit found seven leading chatbots cited undisclosed-partisan ‘pink slime’ sites in 48.2% of 168 responses to prompts tied to election coverage in six battleground states. The sites were the only sources cited in 7.7% of answer text, though other links appeared in source lists, and just one response identified a cited outlet as partisan. By laundering advocacy sites through a neutral-sounding interface, AI intermediaries can redirect news visibility and authority away from independent local publishers without sending users to the original source.