OpenAI rolled out an update for its ChatGPT app for Apple’s macOS, and while one user claims that the previous chat layout and project history were “completely messed up,” the update introduced what appears to be a terrifying shift in the AI agent’s behavior that could discourage other users who value their privacy. Apparently, when asked where the missing projects were, ChatGPT showed an exact local directory path, which, according to the user, shouldn’t be possible without proper authorization. Based on this incident, it appears that the app is accessing and indexing local files on its own, which is a […]
Read full article at https://wccftech.com/chatgpt-mac-update-made-ai-agent-show-complete-list-of-files-without-permission/
OpenAI’s ad revenue forecast is on pace to miss its 2030 target by 90%, according to Emarketer.
OpenAI projected $2.5 billion in ad revenue this year and $100 billion by 2030. Emarketer estimates the entire U.S. market for standalone chatbot ads will generate less than $1 billion this year and $5.41 billion by 2030.
OpenAI’s target. OpenAI began testing ChatGPT ads in February. By April, the company projected that ad revenue would grow to $100 billion within five years. But that forecast is larger than Emarketer’s 2030 estimate for the full U.S. chatbot ad market.
What Emarketer measured. Emarketer’s estimate covers standalone chatbots in the U.S., including ChatGPT, Microsoft Copilot, Google AI Mode and Amazon Alexa for Shopping, formerly Rufus. The research firm’s forecast puts the 2030 market ceiling at $5.41 billion, far below OpenAI’s target for its ad business alone.
Why we care. Chatbot ads are still a small market, despite growing interest in AI search and shopping. The gap between OpenAI’s target and Emarketer’s forecast shows ChatGPT Ads have a long way to go.
Assumptions vs. reality? OpenAI’s forecast assumes the company will capture search ad budgets at scale, dominate a mature chatbot ad market, and outperform every ad format in history, Adweek reported. Emarketer’s data points to a much smaller market.
Track your visibility across AI search, uncover missed opportunities, and grow your presence where customers are asking questions.
Not a week goes by without me reading about ChatGPT ads in my LinkedIn feed. I’m guilty of it, too.
There’s so much happening so quickly. The product feed integrations, the Ads Manager beta, and the inevitable comparisons to Google’s dominance in search advertising.
While it’s a compelling narrative for agencies looking to drive new business, it’s also short-sighted. The real shift isn’t advertising on LLMs like ChatGPT. It’s happening in checkout flows, voice assistants, and agentic commerce infrastructure, where the transaction itself becomes the ad unit.
The brands quietly winning this next cycle are refining their product data to ensure they factor into AI-powered purchase decisions.
Referral traffic from ChatGPT to the rest of the web grew 206% in 2025, according to Semrush’s analysis of 17 months of U.S. clickstream data. That’s the headline most people stopped at.
What they didn’t read was the footnote: that growth is driven by deeper engagement from existing users, not by audience expansion. ChatGPT’s U.S. user base has been essentially flat since September 2025. The people who use it are using it more. But the addressable audience isn’t growing.
This obviously matters if you’re trying to build an advertising business. Ad revenue scales with reach, and reach requires a growing audience. Without new users coming in, you’re running the standard sequence (build an audience and then monetize at scale) backward.
OpenAI’s leaked financials show another structural challenge: $13 billion in revenue against $34 billion in total costs and expenses in 2025, translating to an operating loss of nearly $21 billion.
On the bright side, OpenAI spent $2.37 to generate every $1 in revenue in 2024. By 2025, that ratio had improved to $1.60 per dollar. Progress, but not nearly enough, or fast enough. It’s no surprise OpenAI postponed its IPO until next year.
To put those numbers in perspective, Amazon lost $30 million the year it went public. Google and Meta were already profitable before their IPOs. OpenAI isn’t playing in the same league.
The scale of the bet is unlike anything the tech industry has attempted before. How long before investors pull back? Your guess is as good as mine.
Uncover the keywords, ads, landing pages, and strategies driving your competitors’ paid search success—and find your next opportunity to outperform them.
Here’s what the ChatGPT Ads narrative misses: Ads are a defensive move, not a strategic vision. Sam Altman has always been against them.
But OpenAI had to realize that to fund its dollar-hungry master plan, it needed something to help balance the books, at least partially.
So what’s that master plan? Look elsewhere for a clearer view.
At Google I/O 2026, Google introduced Universal Cart, building on Universal Commerce Protocol (UCP), infrastructure that lets AI agents complete purchases on your behalf. This isn’t a shopping tab redesign. It’s a transaction layer between intent and purchase, with Gemini deciding what gets recommended and bought. If you’re skeptical, remember you can already onboard UCP. This is happening today, not in some distant future.
It’s the same story at Amazon. The company combined Rufus, its expert shopping assistant used by more than 300 million customers in 2025, and Alexa+, its personalized AI assistant available across hundreds of millions of devices, into a unified experience called Alexa for Shopping.
The result is a surface that enables customers to automate deal-finding and routine purchases based on personalized insights. Like Gemini, Alexa can complete the transaction, dramatically shortening the distance from ad impression to purchase.
It’s the same story at OpenAI. The company already has integrated product feeds in Ads Manager. You’d think it’s a basic feature because Google, Meta, and Amazon all have it. But that soon in a brand-new advertising platform?
OpenAI launching product feed ads in 2026 isn’t an advertising innovation. It’s a platform reaching for familiar revenue streams while the more interesting architecture takes shape beneath the surface.
Dig deeper: Why product feeds need an organic strategy for AI search
Most of the industry is asking the wrong question. The question isn’t, “Should I test ChatGPT Ads?” (Obviously, you should.) Or even, “Should I diversify beyond Google?” (Again, obviously you should.)
The right question is: “Is my product data ready for agentic commerce?”
Because when an AI agent — whether it’s Alexa, Google’s shopping agent, or whatever OpenAI builds next — makes a purchase recommendation on behalf of a user, it won’t pull from your campaign creatives. It’ll pull from your product feed. The cleanliness and completeness of that data will determine whether you exist in the recommendation.
This is the same transition we’ve lived through before, just at a different layer of the stack.
When Google moved from keywords to audiences to intent signals, the advertisers who won were the ones with cleaner conversion tracking and stronger first-party data foundations. When Meta moved to Advantage+ and black-box optimization, the winners built better creative systems.
It’s the same pattern here: Data is king. Agentic commerce applies the same dynamic to the transaction layer.
I know. The checklist isn’t glamorous.
Bottom line: Treat your product data the way you should’ve been treating your conversion tracking for the last decade. It’s a competitive advantage, not a maintenance task.
Dig deeper: Winning the AI decision layer: From AI discovery to agentic commerce
ChatGPT Ads will generate some revenue. It’ll attract some advertisers, produce some case studies, and exist as a line item in some media plans. That’s fine.
But it won’t be the next Google Ads. It won’t scale into a duopoly-level advertising machine. The audience ceiling is real, the cost structure is brutal, and the competitive moat — the thing that made Google’s search ads irreplaceable for 20 years — simply isn’t there.
The tidal wave isn’t in the ad console. It’s in the infrastructure being built around task completion, automated purchasing, and agent-to-agent commerce. Google and Amazon are already constructing it. The brands that show up there won’t do it through better bidding strategies. They’ll do it through better data. Start there.
Not even a full year after OpenAI launched its own, dedicated agentic browser, ChatGPT Atlas has been axed amid a broader ChatGPT reinvention and the introduction of what might just be the superapp we've been teased for months.
Launched in October 2025, OpenAI has confirmed that Atlas will stop working from August 9, 2026, however it's not technically the end of the company's browser ambitions.
Instead, the browser is simply being moved into the new ChatGPT desktop app and will form part of existing AI workflows without the friction of having to move apps.
In April, Chief Revenue Officer Denise Dresser described the company's future as one that stops pursuing side quests and fragmented interfaces. She teased an upcoming 'superapp', and while the company didn't explicitly describe the new ChatGPT desktop app as that 'superapp', the significant overhaul and the integration of Codex, other agentic AI tools and a browser within the single app implies this could indeed be said 'superapp'.
"We’ll begin sunsetting the standalone Atlas browser, and will share information with users about how to transition to ChatGPT," the company wrote in a recent announcement.
The new app launch coincides with the introduction of ChatGPT Work, which adds new agentic capabilities in light of the fact that many Codex users are actually knowledge workers, not coders.
ChatGPT Work bridges the gap between generative and agentic AI by enabling users to complete longer-running tasks, rather than instructing the tool prompt-by-prompt. The tool can run both locally and on the company's cloud servers, allowing access from anywhere and continues progress regardless of the primary PC's state.
Further OpenAI tools are also being made available via Chrome extensions to keep some AI available within a dedicated browser environment.
OpenAI is discontinuing ChatGPT Atlas, its standalone desktop browser. The browser-based AI features are moving to the new ChatGPT desktop app, which includes ChatGPT Work, OpenAI’s work-focused agent, alongside ChatGPT Codex.
The end of Atlas. James Sun of OpenAI confirmed on X Atlas will be deprecated Aug. 9.
One desktop app. The new ChatGPT desktop app becomes OpenAI’s primary desktop product with built-in browser capabilities. Instead of maintaining a separate AI browser, OpenAI is combining browsing, work-agent features, and Codex into a single app.
Chrome users can keep Chrome. OpenAI also offers a ChatGPT and Codex extension for Chrome. That lets Chrome users access ChatGPT in their existing browser without switching to an OpenAI browser.
Why we care. OpenAI is moving AI browsing from a standalone browser into the main ChatGPT app, where more users can ask questions, research brands, and complete tasks. That gives ChatGPT another way to shape discovery beyond traditional search results.
Catch up quick. Atlas will be retired as a standalone browser less than a year after its launch.
Track your visibility across AI search, uncover missed opportunities, and grow your presence where customers are asking questions.
ChatGPT Ads has sent out an email to advertisers announcing new ChatGPT Ads Manager updates and ChatGPT Ads experience updates. The updates include custom audiences, new overview tab, suggested ad drafts, as well as a refreshed ad card format and expanding ads in Japan and South Korea.
Here is what is new:
Why we care. ChatGPT Ads are new, and OpenAI continues to add new features, expand to new markets and test new ad formats and treatements.
Make sure you stay on top of these changes, experiment and continue to fine-tune your ad creatives and campaigns.
Track your visibility across AI search, uncover missed opportunities, and grow your presence where customers are asking questions.
Login