cbsnews.com blocks GPTBot
Browser: HTTP 200 · control fetch
| AI crawler | robots.txt | Live HTTP | Verdict |
|---|---|---|---|
| GPTBotOpenAI (ChatGPT) | Disallow: / | HTTP 200 | Blocked |
| ClaudeBotAnthropic (Claude) | — | HTTP 200 | Accessible |
| PerplexityBotPerplexity | — | — | Accessible |
| Google-ExtendedGoogle (Gemini/Vertex) | — | — | Accessible |
| CCBotCommon Crawl (training data) | — | — | Accessible |
| anthropic-aiAnthropic (legacy token) | — | — | Accessible |
robots.txt found
What the check found on cbsnews.com
Measured on 14 November 2026, 04:39 UTC: robots.txt of cbsnews.com read, homepage requested with the GPTBot and ClaudeBot user agent, and with a browser as control (HTTP 200).
- GPTBot: robots.txt blocks GPTBot with “Disallow: /” in a group of its own, “User-agent: GPTBot”. Yet the server still serves it the homepage with HTTP 200, so the block holds only as long as the crawler obeys robots.txt.Collects content that may be used to train OpenAI’s models.
- ClaudeBot: ClaudeBot got HTTP 200 in the live request; robots.txt does not name it.
No rule of their own in robots.txt and no live request: PerplexityBot, Google-Extended, CCBot, anthropic-ai.
Excerpt from the robots.txt of cbsnews.com
The groups that apply to one of the checked AI crawlers, as the check read them.
User-agent: * Disallow: */search/* Disallow: /index.php/* Disallow: /election-results-data/* Disallow: /common Disallow: /htdocs/* Disallow: /user/login* Disallow: /app/?* Disallow: /network Disallow: /assets Disallow: /stories Disallow: /sections Disallow: */feed/ # + 23 more rules User-agent: GPTBot Disallow: /
Verify it yourself
curl -A "GPTBot/1.2" -s -o /dev/null -w "%{http_code}\n" https://cbsnews.comGPTBot can’t load cbsnews.com. OpenAI uses GPTBot for training data; ChatGPT search relies on a separate crawler, OAI-SearchBot, which this check does not test. If the block sits in the hosting layer (CDN, WAF, site builder) and filters bots by user agent, it may hit that crawler as well. Most blocks like this are not a deliberate choice, so they go unnoticed. The fix is usually a single allow-rule; the harder question is whether you’re then actually cited.
Related check
Is cbsnews.com actually cited in ChatGPT?Crawlability is only step one — a reachable site still has to be named in the AI answer to win the customer.