Update, 17 October: OpenAI on 16 October added Slides to ChatGPT: editable, shared presentations in Space that teammates can co-edit, present from ChatGPT or export to PowerPoint, PDF or Google Slides, rolling out to eligible Pro, Business and Enterprise users. Details below.
OpenAI’s ChatGPT account posted on 15 October 2026 that “ChatGPT can now use your tabs in Chrome to make answers more useful and get things done for you.” In Chat, it said, ChatGPT can look across open tabs and gather information; in Work, “ChatGPT can control your Chrome tabs” to send messages, fill out forms or download files.
OpenAI’s documentation says the browser extension works in Google Chrome, Microsoft Edge, Brave, Opera and Vivaldi from the ChatGPT desktop app, and that ChatGPT asks before it interacts with each new website.
Business Engineer Video · AI AGENTS
A Business Engineer explainer on AI agents, software that moves from answering questions to doing tasks. It covers the concept in general and says nothing about OpenAI’s announcement.
The key insight: As we read it, the announcement splits reading from acting: OpenAI says Chat can look across open Chrome tabs, while Work can control them, and its documentation routes that control through the desktop app in Work or Codex chats.
The thread opened: “We’re bringing the power of computer use to anyone who uses Chrome.” A second post said ChatGPT inside a Chrome tab had been upgraded, with the full ChatGPT history available and the option to ask questions in Chat or get things done in Work side by side with the site.
The examples OpenAI posted include comparing three couches by dimensions and return policy, filling a registration form from an event brief and stopping before submitting, and putting vendors from open tabs into a comparison sheet. A final post linked the extension in the Chrome Web Store and said: “Available for all ChatGPT plans.”
Per OpenAI’s browser extension documentation: Chrome, Edge, Brave, Opera and Vivaldi all support tab mentions and browser control from the ChatGPT desktop app; Chrome, Edge, Brave and Vivaldi also support side chat, and Opera does not.
Chat Reads, Work and Codex Act
The posts draw a line between the two modes: Chat can look across open tabs and gather information, while Work can control the tabs. OpenAI’s setup guide says control runs through the desktop app: open Settings, then Computer Use, install the extension, then “Start a ChatGPT Work or Codex chat and select your browser with an” @-mention.
The documentation says that “All five browsers support tab mentions and browser control from the desktop app.” Chrome, Edge, Brave and Vivaldi also support side chat, a ChatGPT panel beside the page; “Opera doesn’t support side chat”, so its tasks start in the desktop app.
OpenAI’s post says the extension is available on all ChatGPT plans. Neither the posts nor the documentation we read say which plans include Work or Codex, and the documentation adds that “Browser availability can depend on rollout and your workspace settings.”
How Website Access Works
According to the documentation, “By default, ChatGPT asks before it interacts with each new website.” The user can choose Allow once, Allow for this site, Allow for all sites or Decline. OpenAI marks the all-sites option as an elevated risk and says to choose it only if you trust ChatGPT to use any website open in the browser.
Allowed and blocked domains sit in an allowlist and a blocklist under Manage, and the supported browsers share these permissions. For browser history, OpenAI says ChatGPT asks each time and that “history doesn’t have an always-allow option.”
What the Extension Can Touch
OpenAI lists the permissions Chrome may ask for at install, including “Access the page debugger”, “Read and change all your data on all websites”, browsing history on signed-in devices, bookmarks, downloads and tab groups. It says ChatGPT still applies its own confirmations, allowlists and blocklists before using websites or history in a task.
On data, the documentation says: “OpenAI doesn’t store a separate complete record of your browser actions from the extension.” Browser activity is stored when it becomes part of the chat context, such as page text, screenshots and tool calls, and the Memories setting applies to browser work.
Update, 17 October: ChatGPT Slides Arrives in Work and Space
OpenAI’s ChatGPT release notes dated 16 October say “ChatGPT Slides lets you create presentations and keep working on them with your team in Space,” starting from a conversation, files or an outline, or from an example deck shared “to guide the style and layout.” Users can edit text, images, shapes and charts directly or ask ChatGPT to make changes, invite teammates to edit the same deck in real time and leave shared comments, and present directly from ChatGPT.
On the web, decks export to PowerPoint, PDF or Google Slides; in the desktop app they open in PowerPoint or Keynote when those apps are installed, the notes say. According to OpenAI’s help article, Slides is available on Pro, Business and Enterprise plans with Space access, users start by asking ChatGPT in Work for a deck or by choosing Create > Slide in Space, and collaborators can @mention ChatGPT in comments to have it make edits. Comment-only access “isn’t available at launch,” the article says. OpenAI did not give pricing or usage figures.
As we read it, this is the same Work surface this piece describes acting in Chrome tabs, now producing the file itself: the deck lives in ChatGPT and leaves as a PowerPoint or Google Slides export, rather than ChatGPT working only inside those apps.
Update, 20 October: ChatGPT Adds PIN-Locked Chats
OpenAI added locked chats to ChatGPT on 20 October, according to its release notes: “Locked chats let you keep selected conversations in your account while hiding them from regular chat history, pinned chats, and search.” They open with a 4-digit account PIN, or on supported mobile devices with Face ID, Touch ID or the device unlock method, its help center says, and on desktop they need the new ChatGPT app for macOS or Windows.
OpenAI’s help article says locked chats are available on ChatGPT Free, Go, Plus, Pro, Business, Enterprise, and Edu, and that they are rolling out gradually; the release note names Free, Go, Plus, and Pro users. They are not available on ChatGPT for Healthcare, and Codex conversations can’t be locked, the article says.
The help article also sets out the limits. Locking a chat does not “create end-to-end encryption or a separate encrypted vault” and does not change how long OpenAI retains the chat. “Resetting your account PIN permanently deletes all locked chats,” it says, and OpenAI Support cannot recover the deleted content.
The Structural Read
As we read it, the signed-in browser becomes the agent’s workplace: OpenAI’s documentation says ChatGPT can read or act on sites where you are already signed in, such as LinkedIn, Salesforce, Gmail or internal tools.
As we read it, permission is the product surface: OpenAI says ChatGPT asks before each new website, offers four answers from Allow once to Decline, and keeps one allowlist and blocklist shared across the supported browsers.
As we read it, the browser choice is wide but the entry point is narrow: OpenAI says all five browsers support browser control from the desktop app, and that availability can depend on rollout and workspace settings.
OpenAI, ChatGPT browser extension documentation
“Treat page content as untrusted context, and review the website before allowing ChatGPT to continue.”
Three Implications
IT AND SECURITY TEAMS OpenAI lists install permissions including access to the page debugger and reading and changing all data on all websites, and says ChatGPT still applies its own confirmations, allowlists and blocklists.
TEAMS USING CHATGPT AT WORK OpenAI says control starts in a Work or Codex chat in the desktop app, with the browser picked by an @-mention, and that the Memories setting applies to browser work.
ANYONE WEIGHING THE ALL-SITES OPTION OpenAI marks Allow for all sites as an elevated risk and says to choose it only if you trust ChatGPT to use any website open in the browser.
The framework argues that value migrates from the model to the harness, and puts it this way: “Pricing power leaves the model and lands on whoever controls the surface and its reusable workflows.”
As we read it, OpenAI’s Chrome extension extends ChatGPT’s harness into the browser where users are already signed in. The framework predates the announcement and does not discuss it.
Per OpenAI’s browser extension documentation: ChatGPT asks before it interacts with each new website, with four answers; OpenAI marks Allow for all sites as an elevated risk. Browser history is asked per request and has no always-allow option.
What Is Not Established
OpenAI’s posts and documentation do not say which ChatGPT plans include the Work mode that controls tabs, and we do not state it. The documentation says availability can depend on rollout and workspace settings.
We read the posts and the documentation in full on 17 October 2026. We did not install or test the extension and did not contact OpenAI.
Business Engineer Framework
The Agentic Harness War
A Business Engineer framework on why value moves from the model to the harness that agents work through.
OpenAI said on 15 October 2026 that ChatGPT can use Chrome tabs: reading across them in Chat and controlling them in Work, with the extension available for all ChatGPT plans.
Its documentation says control runs from the desktop app in Work or Codex chats across Chrome, Edge, Brave, Opera and Vivaldi, with ChatGPT asking before each new website.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. On 17 October 2026 we read in full the ChatGPT account’s posts on X of 15 October 2026 and OpenAI’s browser extension documentation; every quotation and fact in this piece comes from them. We did not install or test the extension and did not contact OpenAI. The references to our earlier reporting link to our own pieces.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
