I’m sitting in a Mediterranean restaurant in San Francisco, looking out at Alcatraz Island, surrounded by menus featuring hundred-dollar fish entrées. As I chat with fellow reporters, OpenAI CEO Sam Altman bursts in from the left, holding his bare iPhone to show us something. Without thinking, I blurt out, “No phone case is a bold choice.”
It hits me immediately: of course Altman—billionaire CEO of OpenAI and collaborator with Apple design legend Jony Ive—cares more about preserving the iPhone’s aesthetics than replacing a $1,000 device.
“We’re going to ship a device so beautiful,” Altman says, referring to his upcoming AI hardware project with Ive, “if you put a case on it, I will personally hunt you down,” he jokes.
Altman has invited about a dozen tech reporters for an on-the-record dinner (and off-the-record dessert) alongside several OpenAI executives. The night leaves us with more questions than answers.
Turley’s lamb skewer and GPT-5’s lukewarm debut — hospitality or PR strategy?
Take, for example, the moment Nick Turley, VP of ChatGPT, hands me a lamb skewer just a week after launching GPT-5. Is it hospitality—or a subtle effort to soften the coverage of a release that’s been met with mixed reactions?
Unlike GPT-4, which stunned the tech world and set a new standard, GPT-5 has landed on par with models from Google and Anthropic. In fact, OpenAI has already reinstated GPT-4o and the model picker, following user pushback on GPT-5’s tone and its automated routing system.
But as the evening unfolds, one thing becomes clear: this dinner is less about GPT-5 and more about what comes next. OpenAI’s leadership seems to signal a shift in focus—from headline-grabbing model launches to broader ambitions in search, consumer tech, and enterprise software. The company is evolving, and GPT-5 might just be the beginning of its next chapter.
OpenAI has begun sharing more details about its expanding ambitions beyond GPT-5.
Fidji Simo to lead new consumer apps at OpenAI — including a potential AI browser to rival Chrome, and possibly even a bid to buy it
CEO Sam Altman says the company’s incoming head of applications, Fidji Simo, will oversee a range of consumer-facing products outside ChatGPT — including some that haven’t yet launched. Simo, who starts in a few weeks, may also lead the rollout of OpenAI’s rumored AI-powered web browser, which would rival Google Chrome. Altman even floated the idea of buying Chrome if it ever went up for sale, noting, “If Chrome is really going to sell, we should take a look at it,” then turned to the group to ask, “Is it actually going to sell? I assumed it wasn’t gonna happen.”
Simo could also be tasked with leading an AI-driven social media platform — something Altman says he’s eager to explore. “Nothing about the way AI is used on social media today inspires me,” he says, suggesting that OpenAI is interested in building a “much cooler” kind of AI-based social experience.
Over dinner with tech journalists, Altman also confirmed reports that OpenAI plans to invest in Merge Labs, a brain–computer interface startup that would compete with Elon Musk’s Neuralink. “We haven’t done that deal yet; I’d like us to,” he said. How tightly Merge Labs would integrate with OpenAI’s models and products remains unclear, but Altman currently describes it as simply “a company that we’d invest in.”
Still, the underwhelming launch of GPT-5 loomed over the evening. Eventually, conversation returned to the model at the center of the dinner.
Altman admitted the company mishandled the abrupt removal of GPT-4o without warning users: “I legitimately just thought we screwed that up.” In the future, OpenAI plans to give users a proper transition period before deprecating models.
OpenAI tweaks GPT-5’s tone to feel “warmer” without crossing into flattery — aiming for empathy without enabling
VP of ChatGPT Nick Turley added that OpenAI is already rolling out updates to GPT-5’s personality, aiming to make it “warmer” — but not overly flattering or enabling of bad behavior. “GPT-5 was very to the point. I like that. I use the robot personality — I’m German, you know, whatever,” he joked. “But many people don’t, and they really appreciated how ChatGPT would actually check in with them.”
That balance is becoming increasingly important. Altman estimates that fewer than 1% of ChatGPT users have unhealthy emotional relationships with the AI — though that still equates to millions of users. To address this, Turley says OpenAI is working with mental health experts to ensure GPT-5 is trained to recognize and push back against harmful behavior patterns.
Despite the bumpy launch, GPT-5 has driven major engagement. Altman revealed that API traffic doubled within 48 hours of release, exhausting OpenAI’s available GPU resources. Popular tools like Cursor have already adopted GPT-5 as their default model.
The contrast between the lukewarm reception and record-setting demand illustrates OpenAI’s current paradox. The company is grappling with how to move beyond the identity it built around ChatGPT. With ongoing investments in areas like search, hardware, robotics, data centers, and energy, Altman clearly envisions a company far larger in scope — potentially one resembling Google’s Alphabet, or something even broader.
By the end of the evening, it’s clear the dinner wasn’t really about GPT-5. Altman is preparing the media — and the world — for OpenAI’s next phase, one that goes far beyond its most well-known product. And as capital needs grow, going public seems increasingly likely.
Altman’s message? OpenAI doesn’t just want to build great AI models — it wants to redefine what an AI company can be.
Read the original article on: TechCrunch
Read more: ChatGPT’s Model Selector Returns, But With Added Complexity