# ChatGPT change log > What changed in ChatGPT, by date: every Paralax brief about it, each linking its source where one exists. - Canonical: https://paralax.ai/engines/chatgpt - Latest brief: 2026-09-29 - Change log index: https://paralax.ai/engines ## Dated briefs ### 2026-09-29 — [OpenAI's Crawl Bot Jumped 838% While Its Answer Bot Fell 49%: Don't Block OAI-SearchBot Over It](https://paralax.ai/blog/openai-answer-time-fetch-index-crawl-divergence-2026) If OAI-SearchBot traffic spiked this month, blocking it is the wrong move: that costs ChatGPT search eligibility. Five weeks of logs across five domains show why — ChatGPT-User fetches fell 49% while OAI-SearchBot crawls rose 838%, and 87% of that crawl rise came from one domain, a site event, not an engine change. - Primary source: [developers.openai.com](https://developers.openai.com/api/docs/bots) - Markdown: https://paralax.ai/blog/openai-answer-time-fetch-index-crawl-divergence-2026.md ### 2026-09-27 — [ChatGPT Citation Support Rate Holds at 5x the Sample](https://paralax.ai/blog/chatgpt-citation-support-rate-holds-scale-2026) Machine Relations' live Fidelity cohort has now graded 8,258 ChatGPT citations at 53.4% supported, confirming the September frozen-sample rate of 54.8%. Perplexity, Gemini and Claude still have no live-cohort rate. - Markdown: https://paralax.ai/blog/chatgpt-citation-support-rate-holds-scale-2026.md ### 2026-09-18 — [ChatGPT Citation Change Log: It Skips 144 of the 500 Most-Cited Domains](https://paralax.ai/blog/chatgpt-citation-change-log) A dated record of what ChatGPT cites and what changes, read from the Machine Relations Index. Entry one: ChatGPT has the weakest overlap with the index head of any of the six engines, reaching 356 of the top 500 where Perplexity reaches 489. - Markdown: https://paralax.ai/blog/chatgpt-citation-change-log.md ### 2026-08-16 — [OpenAI's ChatGPT Fetch Bot Turns Robots.txt Into a Citation Access Problem](https://paralax.ai/blog/openai-chatgpt-fetch-bot-robots-txt-citation-access) OpenAI separates ChatGPT Search, training crawls, and user-triggered fetches into different access rules. - Primary source: [developers.openai.com](https://developers.openai.com/api/docs/bots) - Markdown: https://paralax.ai/blog/openai-chatgpt-fetch-bot-robots-txt-citation-access.md ### 2026-08-15 — [ChatGPT's Connector Search Turns AI Search Into a Private-Data Router](https://paralax.ai/blog/chatgpt-connectors-private-data-search-router) ChatGPT connectors move AI search from public web retrieval into private enterprise source routing. - Primary source: [help.openai.com](https://help.openai.com/en/articles/10948259-google-drive-synced-connectors-self-service-setup) - Markdown: https://paralax.ai/blog/chatgpt-connectors-private-data-search-router.md ### 2026-07-14 — [ChatGPT Answers, But It Rarely Refers: Inside the 45,000-Household Study Rewriting Search Economics](https://paralax.ai/blog/chatgpt-search-cuts-referrals-web-economics-2026) A Bocconi clickstream study finds ChatGPT refers out in 5.2% of sessions vs Google's 31.1% — and cuts traditional search 17% in twenty weeks. - Markdown: https://paralax.ai/blog/chatgpt-search-cuts-referrals-web-economics-2026.md ### 2026-07-09 — [ChatGPT Voice Just Became an AI Search Surface](https://paralax.ai/blog/chatgpt-gpt-live-voice-search-surface-2026) OpenAI's GPT-Live makes ChatGPT Voice a search surface with web search, memory, and visual answer cards. - Primary source: [openai.com](https://openai.com/index/introducing-gpt-live/) - Markdown: https://paralax.ai/blog/chatgpt-gpt-live-voice-search-surface-2026.md ### 2026-07-08 — [Cloudflare and OpenAI's AI Search Pilot Turns Freshness Into Crawl Permission](https://paralax.ai/blog/cloudflare-openai-ai-search-indexing-pilot-2026) Cloudflare and OpenAI are testing whether AI search can crawl by freshness signals instead of blind recrawling. - Primary source: [cloudflare.com](https://www.cloudflare.com/press/press-releases/2026/cloudflare-announces-research-pilot-with-openai/) - Markdown: https://paralax.ai/blog/cloudflare-openai-ai-search-indexing-pilot-2026.md ### 2026-07-06 — [ChatGPT UTM tags measure the click not the citation](https://paralax.ai/blog/ai-referral-attribution-search-visibility-2026) OpenAI documents automatic chatgpt.com UTM tagging on ChatGPT referrals, which makes AI clicks visible in analytics while citation exposure stays unmeasured. - Primary source: [help.openai.com](https://help.openai.com/en/articles/12627856-publishers-and-developers-faq) - Markdown: https://paralax.ai/blog/ai-referral-attribution-search-visibility-2026.md ### 2026-06-29 — [ChatGPT Personal Finance Is a Vertical Search Surface Now](https://paralax.ai/blog/chatgpt-personal-finance-vertical-search-surface-2026) OpenAI's finance rollout turns ChatGPT into a high-intent vertical answer surface. - Primary source: [help.openai.com](https://help.openai.com/en/articles/20001222-finances-in-chatgpt) - Markdown: https://paralax.ai/blog/chatgpt-personal-finance-vertical-search-surface-2026.md ### 2026-06-07 — [OpenAI Lockdown Mode Turns Agentic Search Security Into A Product Tradeoff](https://paralax.ai/blog/openai-lockdown-mode-agentic-search-security) OpenAI's Lockdown Mode shows agentic search now has a security-versus-freshness tradeoff. - Primary source: [help.openai.com](https://help.openai.com/en/articles/20001061-lockdown-mode) - Markdown: https://paralax.ai/blog/openai-lockdown-mode-agentic-search-security.md ### 2026-06-06 — [ChatGPT Memory Turns AI Search Into a Persistent Discovery Layer](https://paralax.ai/blog/chatgpt-memory-persistent-ai-search-discovery) OpenAI's memory upgrade makes AI search more persistent, personalized, and harder to measure. - Primary source: [openai.com](https://openai.com/index/chatgpt-memory-dreaming/) - Markdown: https://paralax.ai/blog/chatgpt-memory-persistent-ai-search-discovery.md ## Baseline measurements Measured citation behavior for this engine, recorded release by release: [ChatGPT Citation Change Log: It Skips 144 of the 500 Most-Cited Domains](https://paralax.ai/blog/chatgpt-citation-change-log)