ChatGPT search switched en masse to the site: operator
On August 20, Simon Willison relayed a measurement from Promptwatch showing that ChatGPT abruptly began narrowing its searches to preselected domains. The share of queries carrying the site: operator held between 0.3% and 0.5% for weeks, then jumped to 16% to 17% on August 8 — two days after OpenAI announced accuracy fixes in GPT-5.6. Willison concludes that the search tool internally has the shape search(query, recency, domains), which means the model picks domains before it even frames the question. If you care about visibility in that search, you are no longer competing only for a place in the result but for your domain to make the list the model assembles on its own.
Source
ChatGPT search now uses the site:operator at scaleSimon Willison’s Weblog · Original published August 20, 2026
On August 20, Simon Willison drew attention to a measurement showing that the behavior of ChatGPT-driven search changed abruptly in early August. Instead of searching broadly and then choosing sources, the model now often narrows to specific domains up front. For anyone who cares about their site's visibility, that changes the question worth asking.
What was measured
Promptwatch, a tool that tracks how conversational systems assemble answers, measured the share of queries in which the site: operator appears. That share held between 0.3% and 0.5% for weeks, then jumped to 16% to 17% on August 8. The jump comes two days after OpenAI announced fixes to factual reliability in GPT-5.6.
What the tool looks like from the inside
Willison concludes that the model is not being prompted to type the operator itself, but that the search tool has the shape search(query, recency, domains). Domains are therefore a separate parameter of the call, as much as the query is. The practical consequence is that the model decides where to look before it formulates what it is looking for.
What that means for your site
If search narrows to a list of domains, visibility stops being only a question of position in the results. The first question becomes whether the model considers your domain worth writing into the domains parameter at all, and that depends on how strongly your site is tied to a subject rather than on any single page. A drop in citations from Reddit was recorded alongside it, which suggests the ordering of sources in answers changes faster than the ordering in classic search ever did.
What is not confirmed yet
Promptwatch is a commercial tool, and its measurement methodology is not publicly verifiable. OpenAI has confirmed neither the change in tool behavior nor the shape of the call; the conclusion about a domains parameter is Willison's reading, inferred from observed behavior. Read the figures as an indication of direction, not as published data.
Related
llm 0.33 takes the pin out and moves to httpx2
Version 0.32.1 held fresh installs together by pinning the OpenAI Python package below 3.0.0. That was a holding action, and 0.33 does the actual repair: it upgrades to the OpenAI library 3.x and switches its HTTP client from httpx to httpx2. The release also carries changes you will notice in daily use, including a --key option on the embedding commands and server-side tool results that finally show up in llm logs.
Simon Willisonverified
A fresh install of llm broke because the OpenAI library stopped using httpx
Version 0.32.1 pins the OpenAI Python package below 3.0.0 so that new installations work again. Nothing in llm changed to cause the break: it imported httpx while relying on the OpenAI package to bring it along, and when that package dropped httpx the dependency simply stopped arriving. It is a small release with a lesson that outlives it.
Simon Willisonverified
Next.js will patch one critical vulnerability on August 26
Vercel published the date before the fix. Versions 16.3.3 and 15.5.24 arrive on August 26 with a full advisory, and the announcement exists so teams can schedule the upgrade instead of discovering it from a CVE feed.
Next.js Blogverified