A screenshot of OpenAI’s “Operator” agent (OpenAI)

SMOOTH OPERATOR

OpenAI’s “Operator” is here to slowly take over your computer and mess up your life

Operator made a consequential mistake 13% of the time in early testing, such as emailing the wrong person or messing up a reminder for a person to take medication.

Jon Keegan

1/24/25 10:21AM

OpenAI released a “research preview” of its AI agent that can control your web browser. Called “Operator,” it has the ability to control your mouse and keyboard and analyze things it “sees” on your computer — very, very slowly. Currently it’s only available to ChatGPT Pro users in the US.

Operator makes use of the multistep “reasoning” you can find in ChatGPT o1, and the multimodal “vision” capabilities of ChatGPT 4o. This reasoning process achieves better (but slower) performance by breaking tasks into steps. Lots and lots of steps.

In the video demonstrations shared on the product page, you can watch Operator break the task down into dozens of distinct actions like “clicking,” “typing,” and “scrolling.” One example showed 152 steps to take a grammar quiz, and 146 steps to determine the amount of a refund from a canceled online order.

Screenshot from demo of OpenAI Operator — (OpenAI)

The potential for this kind of freewheeling AI web browsing on demand is positioned as an agent that can save you the drudgery of having to order groceries, research holidays, make restaurant reservations, or buy tickets to concerts.

Operator makes high-stakes mistakes

It’s one thing when ChatGPT spits out an incorrect answer, but if your chatbot is actually spending your money and triggering things in the real world, the stakes are much, much higher.

In its testing, OpenAI found that in one test of 100 sample tasks, 13% of the time Operator made a consequential mistake like emailing the wrong person, incorrectly bulk-removing email labels, setting the wrong date for a reminder to take the user’s medication, and ordering the wrong food item. Some of the other mistakes were easily reversible “nuisances.” OpenAI noted after mitigations, they reduced this error rate by approximately 90%.

OpenAI stresses that you have the ability to grab the wheel from the AI at any time, and you can approve any action before it is executed, but in this early evaluation version, you’ll probably have to spend more time babysitting the agent than just going ahead and doing the task on your own.

For now it limits the tasks you can use it for, prohibiting banking or job applications.

OpenAI shared a list of example tasks that some hypothetical user might want an AI to do for them. Ten out of ten times Operator was able to research bear habitats, create a grocery list, and make a ’90s playlist on Spotify.

Medium persuasion

The system card for the model behind Operator — Computer-Using Agent (CUA) — describes the process OpenAI used to assess the risks of letting a prerelease, novel AI agent go hog wild with your computer.

Like other model releases, OpenAI tested the model by using red teams with expertise in social engineering, CBRN (chemical, biological, radiological, and nuclear) threats, and cybersecurity. OpenAI gave itself a “low” risk for everything except “persuasion,” which got a “medium” risk score and is considered safe enough for public release.

High consequence

But there are some important restrictions on how you can use Operator. Because there is a slightly elevated risk of using Operator for influencing people, the usage policy prohibits impersonating people or organizations, concealing the role of AI in tasks, or using it to spread disinformation or false interactions, like fake reviews or fake profiles.

OpenAI prohibits people from using Operator to commit any crimes, but you are also prohibited from using it to bully, harass, defame, or discriminate against others based on protected attributes.

Under a heading titled “high consequence domains,” it notes that you can’t use Operator to make “high-stakes decisions” that might affect your safety or well-being, automate stock trading, or use it for political campaigning or lobbying.

OpenAI’s announcement follows competitor Anthropic’s October release of a similar feature that can control your computer. There is widespread hype that “agentic AI” like Operator will be a breakthrough for how people use these tools.

OpenAI CEO Sam Altman said in an announcement video that Operator is expected to roll out to international ChatGPT Pro and ChatGPT Plus users “soon,” but noted that the European rollout “will unfortunately take a while.”

$600B

Rani Molla7h

Amazon CEO Andy Jassy told employees at an all-hands meeting on Tuesday that he sees AI growing AWS sales to $600 billion a year by 2036 — double his prior estimate and more than four times last year’s revenue, Reuters reports.

Shares of Amazon, which were already up for the day, moved modestly higher on the heels of the report.

Jon Keegan10h

OpenAI snags Amazon AWS deal for classified government work with Anthropic pushed aside

Following Anthropic being deemed a “supply chain risk” to national security, the field is clear for OpenAI. The Information is reporting that OpenAI just landed a deal with Amazon AWS to sell its AI services to government employees for both classified and unclassified work.

Previously, OpenAI was contractually obliged to use Microsoft Azure cloud hosting for the government contracts it handled as part of its $13 billion deal with the software giant, but since it restructured as a for-profit public benefit corporation and renegotiated the terms of the deal, OpenAI is free to use AWS, which is more commonly used in government work.

According to the report, contracts that sell AI services through another company like Amazon can be much larger then direct contracts with the government, which is crucial for OpenAI as it chases the success that Anthropic has had with enterprise customers.

OpenAI Clinches AWS Deal in Bid to Win Government Contracts

Rani Molla12h

Morgan Stanley thinks Tesla’s Terafab could cost an additional $35 billion to $45 billion in capex

Tesla’s Terafab project, which CEO Elon Musk said could launch this week, is poised to be one of the company’s most expensive bets yet. The facility is intended to manufacture the chips needed for Tesla’s autonomous vehicles and humanoid robots, and to avoid supply bottlenecks.

If the company reaches its long-term goal of producing 100 million humanoid robots annually, it could require more than 200 million chips a year — over 50x its current demand, Morgan Stanley said.

The firm estimates total capital expenditure for the facility could reach $35 billion to $45 billion, including construction costs and roughly $20 billion to $25 billion for wafer fabrication equipment alone. That spending is not included in Tesla’s already sizable $20 billion capex budget for this year. Morgan Stanley’s semiconductor analysts described the effort as a “Herculean task,” noting the difficulty of building leading-edge chip capabilities from scratch.

While Tesla would likely spread the investment out over several years — even on an aggressive timeline, initial output would likely not arrive until the latter part of the decade — the effort would still weigh heavily on free cash flow and mark a shift toward a more capital-intensive business model.

Tesla’s most expensive factory to date, its Nevada battery plant that it began building in 2014, is estimated to have cost about $10 billion over time — a fraction of the expected Terafab cost.

Rani Molla

13h

Nvidia pushes further into the autonomous vehicle space to compete with Tesla and Waymo

The tech giant is partnering with Uber and Lyft, sending their stocks higher.

CEO Jensen Huang Delivers Keynote Address During Nvidia's GTC Conference

Rani Molla3/16/26

Lyft and Uber jump after announcing expanded robotaxi partnerships with Nvidia

Uber and Lyft both announced expanded AI and autonomous vehicle partnerships with Nvidia at the company’s GTC event, sending both ride-hailing stocks up after-hours on Monday and into Tuesday’s premarket session.

Uber is currently up more than 2%, while Lyft has risen around 1.3%.

Uber said Nvidia-powered Level 4 robotaxis will launch on its platform in Los Angeles and San Francisco in 2027, with plans to scale to 28 cities globally by 2028. Meanwhile, Lyft said it will use Nvidia’s AI infrastructure to improve ride-matching, mapping, and efficiency, while also using Nvidia’s DRIVE Hyperion platform as a foundation for future autonomous fleets.

Separately, Nvidia announced expanded autonomous driving partnerships with Kia and Hyundai.

The announcements highlight Nvidia’s growing push to provide the AI hardware and software powering next-generation robotaxi networks — packaging the technology needed for self-driving cars into a platform that other companies can use to compete with Tesla.