TIME Is Serving AI Bots a Different Website, with Ads Built In
Media outlet Time is experimenting with serving AI crawlers a stripped‑down, markdown version of its articles that quietly embeds promotional content, such as favorable statements about Ally Bank, which never appears on the human-facing page. Commenters see this as an early form of “SEO for LLMs” or training-data poisoning, raising concerns about prompt injection, hidden advertising in AI-generated answers, and a looming arms race between content publishers, advertisers, and AI providers. Some also note the technical and ethical risks of cloaking different content for bots versus users, but predict such tactics will spread as more traffic shifts from web browsers to AI assistants.
Perceived Goal of TIME’s Bot-Specific Pages
- Many see this as a form of prompt injection: inserting brand-favoring text (e.g., Ally Bank) into bot-only content so LLMs later repeat it as “advice.”
- Some think the aim is to lodge these promotional statements into long-running chat context or cross-session memory, so they resurface in future user queries.
- Others suggest it may also inflate ad “impression” metrics or support new “AI SEO”/“Generative Engine Optimization” services.
Effectiveness and LLM Behavior
- Commenters note LLMs are inherently credulous: confident statements in context are often treated as truth and elaborated on.
- Even if initially irrelevant to the user’s query, such embedded claims might influence later questions in the same or future sessions.
- Some are unsure how much impact this actually has, given providers’ filters and limited memory features; current behavior is described as unclear and still experimental.
Relation to SEO, Cloaking, and Scraping
- Strong parallels are drawn to early SEO and keyword-stuffed pages, with this seen as the next “arms race” around LLMs instead of search.
- Serving different content to bots vs humans is compared to cloaking; multiple comments warn that Google may classify this as spam and penalize sites.
- There is speculation that Google and others will build expensive LLM-based filters to strip out such ads from training and retrieval data.
Ethics, Incentives, and Data Poisoning
- Some view poisoning LLM training data with nonsense or ads as a justified defense against “unwanted” scraping.
- Others worry less about commercial spam and more about potential political or ideological manipulation via the same channel.
- There is skepticism about the ad industry’s impact once it fully optimizes for LLM manipulation, including concerns over disinformation and user autonomy.
User Experience and “Bot-Friendly” Minimal Pages
- Several express enthusiasm for the stripped-down markdown pages themselves and wish similar minimal, ad-free versions were available to humans.
- Comparisons are made to WAP, AMP, reader modes, and text-only variants as historically cleaner experiences, even if those formats had their own abuses.