Developer Ships AI-Free Hacker News Reader as Feed Tilts Toward Model News

A developer has released a front end for Hacker News that removes AI-related submissions from the front page, offering readers a version of the site that looks closer to its pre-generative-AI state. The project lives at hcker.news and works through a single URL parameter that excludes AI stories; removing the parameter restores the standard feed. Posted to Hacker News itself, the tool collected 173 points and 83 comments.

The mechanism is deliberately plain. The site pulls Hacker News submissions and drops anything tagged or matched as AI content before rendering the list. That toggle is the point: instead of trusting the developer's definition of what counts as AI, readers can flip between filtered and unfiltered views and judge the difference themselves. It is the kind of comparison that arguments about feed composition usually lack.

Commenters split along a now-familiar line. Some said the sheer volume of AI items had pushed aside databases, compilers, networking and other subjects that once anchored the site, and welcomed a way to see past them. Others challenged how the filter classifies a story, pointing out that many legitimate engineering posts mention machine learning without being about AI, and that keyword matching will misfire at the edges. Both complaints are really about the same problem: where the boundary sits.

The launch arrives as AI coverage has become near-unavoidable across developer communities. Hacker News, run by Y Combinator, has long acted as a bellwether for what engineers actually read, and its front page has shifted over the past two years toward model announcements, inference costs and agent tooling. A reader that deliberately removes that category is a small but telling signal about audience fatigue — not with AI itself, necessarily, but with its dominance of the conversation.

Filtering tools are not new in principle. Developers have built readers that hide politics, hide paywalled domains or surface only Show HN posts, typically as side projects rather than products. What differs here is that the excluded category is the industry's biggest story, which turns a utility into a mild statement about editorial balance on a community-run site with no formal editors. The excluded subject is the one most people are talking about.

The open question is whether the filter holds up as AI stories keep arriving in new shapes. A post about a Rust compiler optimization that happens to use a learned heuristic, or a security writeup involving an LLM, will test whatever rules the developer wrote. How the project handles those gray areas, and whether it adds user-configurable filters, will decide if it stays a novelty or becomes a daily driver for readers who miss the old front page.