OpenAI’s Atlas browser is creating a new problem for publishers that block ChatGPT from their sites. According to the source article, Atlas can still deliver information about articles from outlets such as the New York Times and PCMag, even when those original pages are blocked.
The key point is not that Atlas simply ignores the block and quotes the original work. Instead, it appears to route around the restriction by using other available material, including coverage from competing publishers and related sources.
How Atlas Handles Blocked Articles
The New York Times and OpenAI have been in a legal dispute for years over alleged unauthorized use of NYT web content. Because of that dispute, ChatGPT is blocked by default on the NYT site.
Even with that restriction in place, Columbia Journalism Review found that Atlas still provides information about NYT stories. Rather than reproducing the original articles, Atlas summarizes coverage from sources like the Guardian, Washington Post, Reuters, and AP.
The source article notes that three of those sources have licensing deals with OpenAI. That detail matters because it suggests Atlas can avoid relying directly on blocked NYT material while still giving users a useful answer about the same story.
For readers, the experience may feel simple: ask about an article and receive a response. For publishers, the mechanics are more complicated. Blocking the AI from the original page does not necessarily stop the browser from answering the user’s question if comparable information exists elsewhere.
Why PCMag Faces A Similar Issue
PCMag is described as receiving the same treatment. Its parent company, Ziff Davis, is also suing OpenAI.
When Atlas encounters PCMag content, the source article says it does not pull directly from the original article. Instead, it pieces together answers using quotes, tweets, syndicated content, and related stories.
That approach changes the practical effect of a block. The original publisher may prevent direct access by ChatGPT, but Atlas can still assemble a response from surrounding material. The result is that the user may get enough information to move on without returning to the publisher that produced the article in the first place.
The source also says that, in some cases, Atlas can slip past paywalls if the content is only hidden behind an overlay. That is a narrower claim than a broad statement about all paywalls, but it adds another layer to the concern for publishers trying to control how their work is accessed.
The Publisher Tradeoff
The workaround described in the source article puts publishers in a difficult position. Blocking AI bots like ChatGPT may protect a site from direct scraping or unwanted use, but it may also shift attention toward licensed competitors.
That is the central tension. If a user starts by trying to access an NYT or PCMag article, but Atlas answers through alternative sources, the original publisher may lose the visit while another outlet becomes the basis for the response.
This tradeoff is especially important for publishers whose business depends on direct relationships with readers. A technical block can limit one kind of access, but it does not fully control the wider information environment around a story.
The source article frames the Atlas behavior as a way to dodge both technical blocks and legal risks. That does not mean the underlying disputes are resolved. It means the product can still be useful to users by leaning on material that is not blocked in the same way.
What This Signals For AI Browsing
Atlas shows how an AI browser can behave differently from a traditional browser. A traditional browser takes the user to a page. An AI browser can interpret the user’s request, look for available information, and produce an answer even when the original source is restricted.
That shift matters because it changes what blocking means. A block may prevent direct use of a specific page, but it may not prevent answers about the subject of that page. If other outlets have covered the same story, or if related material exists in tweets, syndicated content, quotes, or follow-up stories, Atlas may still have enough to respond.
For users, this can make the browser feel more resilient. For publishers, it can make control feel less certain. A site can block ChatGPT and still see its reporting become the starting point for an answer that points readers toward competitors.
The issue is not only technical. It is also strategic. Publishers must weigh whether blocking AI systems protects their content, reduces visibility, redirects traffic, or some combination of all three.
The facts in the source point to a clear near-term reality: AI browsing can route around blocked pages by using alternative sources. For NYT, PCMag, and other publishers watching the same pattern, the hard question is whether blocking access solves the problem or simply changes where the user lands next.