عربي English Türkçe اردو فارسى
Logo
Latest News
Technology

OpenAI Says Ousted CEO Sam Altman to Return to Company behind ChatGPT

OpenAI CEO Sam Altman testifies before a Senate Judiciary Privacy, Technology & the Law Subcommittee hearing titled 'Oversight of A.I.: Rules for Artificial Intelligence' on Capitol Hill in Washington, US, May 16, 2023. REUTERS/Elizabeth Frantz/File photo
OpenAI CEO Sam Altman testifies before a Senate Judiciary Privacy, Technology & the Law Subcommittee hearing titled 'Oversight of A.I.: Rules for Artificial Intelligence' on Capitol Hill in Washington, US, May 16, 2023. REUTERS/Elizabeth Frantz/File photo
TT
TT

OpenAI Says Ousted CEO Sam Altman to Return to Company behind ChatGPT

OpenAI CEO Sam Altman testifies before a Senate Judiciary Privacy, Technology & the Law Subcommittee hearing titled 'Oversight of A.I.: Rules for Artificial Intelligence' on Capitol Hill in Washington, US, May 16, 2023. REUTERS/Elizabeth Frantz/File photo
OpenAI CEO Sam Altman testifies before a Senate Judiciary Privacy, Technology & the Law Subcommittee hearing titled 'Oversight of A.I.: Rules for Artificial Intelligence' on Capitol Hill in Washington, US, May 16, 2023. REUTERS/Elizabeth Frantz/File photo

The ousted leader of ChatGPT-maker OpenAI is returning to the company that fired him late last week, the latest in a saga that has shocked the artificial intelligence industry.
San Francisco-based OpenAI said in a statement late Tuesday: “We have reached an agreement in principle for Sam Altman to return to OpenAI as CEO with a new initial board" made of former Salesforce co-CEO Bret Taylor, former US Treasury Secretary Larry Summers and Quora CEO Adam D’Angelo.
OpenAI’s previous board of directors, which included D'Angelo, had refused to give specific reasons for why it fired Altman on Friday, leading to a weekend of internal conflict at the company and growing outside pressure from the startup's investors.
Microsoft, which has invested billions of dollars in OpenAI and has rights to its technology, quickly moved to hire Altman on Monday, as well as another co-founder and former president, Greg Brockman, who had quit in protest after Altman's removal. That emboldened a threatened exodus of nearly all of the startup's 770 employees who signed a letter calling for the board's resignation and Altman's return.
Microsoft Chief Technology Officer Kevin Scott put out a call to the startup’s employees Tuesday on X, formerly Twitter: “Know that if needed, you have a role at Microsoft that matches your compensation and advances our collective mission.”
Microsoft CEO Satya Nadella also made clear in a series of interviews Monday that he was still open to the possibility of Altman returning to OpenAI, so long as the startup's governance and board problems are solved.
“We are encouraged by the changes to the OpenAI board,” Nadella posted on X late Tuesday. “We believe this is a first essential step on a path to more stable, well-informed, and effective governance.”
In his own post, Altman said that “with the new board and (with) Satya's support, I'm looking forward to returning to OpenAI, and building on our strong partnership with (Microsoft)."
Co-founded by Altman as a nonprofit with a mission to safely build so-called artificial general intelligence that outperforms humans and benefits humanity, OpenAI later became a for-profit business but one still run by its nonprofit board of directors. It's not clear yet if the board's structure will change with its newly appointed members.
Nadella said Brockman, who was OpenAI's board chairman until Altman's firing, will also have a key role to play in ensuring OpenAI “continues to thrive and build on its mission.”
Hours earlier, Brockman returned to social media as if it were business as usual, touting a feature called ChatGPT Voice that was rolling out for free to everyone who uses the chatbot.
“Give it a try — totally changes the ChatGPT experience,” Brockman wrote, flagging a post from OpenAI's main X account that featured a demonstration of the technology playfully winking at recent turmoil.
“It’s been a long night for the team and we’re hungry. How many 16-inch pizzas should I order for 778 people,” the person asks, using the number of people who work at OpenAI. ChatGPT's synthetic voice responded by recommending around 195 pizzas, ensuring everyone gets three slices.
As for OpenAI's short-lived interim CEO Emmett Shear, the second interim CEO in the days since Altman's ouster, he posted on X that he was “deeply pleased by this result, after (tilde)72 very intense hours of work.”
“Coming into OpenAI, I wasn't sure what the right path would be,” Shear wrote. “This was the pathway that maximized safety alongside doing right by all stakeholders involved. I'm glad to have been a part of the solution.”

Most Viewed

 
 
Technology

Data Center Next Door Divides Old Massachusetts Mill Town

This aerial photo shows the Markley Lowell Data Center surrounded by residential neighborhood homes and apartment buildings in Lowell, Massachusetts, on October 15, 2026. (Photo by Lauren Owens Lambert / AFP)
This aerial photo shows the Markley Lowell Data Center surrounded by residential neighborhood homes and apartment buildings in Lowell, Massachusetts, on October 15, 2026. (Photo by Lauren Owens Lambert / AFP)
TT
TT

Data Center Next Door Divides Old Massachusetts Mill Town

This aerial photo shows the Markley Lowell Data Center surrounded by residential neighborhood homes and apartment buildings in Lowell, Massachusetts, on October 15, 2026. (Photo by Lauren Owens Lambert / AFP)
This aerial photo shows the Markley Lowell Data Center surrounded by residential neighborhood homes and apartment buildings in Lowell, Massachusetts, on October 15, 2026. (Photo by Lauren Owens Lambert / AFP)

Retiree Eileen Castle hasn't been able to enjoy her backyard since a data center moved in next door in Lowell, Massachusetts, bringing noise, fumes, dust and water mist.

But a few streets away, contractor Louis Haas is grateful to Markley, the company running the facility, for redeveloping an abandoned site he says was rife with drug use.

Dubbed "the Dungeon" by some locals for the imposing black silhouette it casts over a residential neighborhood of this old milling city, those opposing views encapsulate how data centers are polarizing communities across the United States.

Republican-governed states, with their looser regulations, host most of the country's current AI build-out. But in liberal-leaning New England, some feel betrayed by their elected representatives, while others fear being left behind if the industry is curbed.

- 'Nothing but noise' -

Over the din of construction behind a spiked fence, Castle, 82, told AFP her great-great-grandparents bought the property.

But since the data center moved into a disused pasta factory a decade ago, "We have had nothing but noise.

We have dust, dirt. We have all kinds of mist" from the adjoining cooling towers, which she fears could carry Legionnaires' disease.

Noxious fumes made a long interview outdoors impossible, and she lamented that her grandchildren can no longer swim in her small pool.

"I think every once in a while that maybe I should sell my house," she says, but adds her property value has plummeted.

"The Markley Group serves more than 600 clients in Massachusetts and connects the digital infrastructure that powers our economy, our schools, and the everyday needs of our community," Markley Group CEO Jeff Markley said in a statement to AFP.

The company said its backup diesel generators run only for five minute tests each week and, thanks to abatement efforts, produce "about 44 decibels of sound, quieter than a soft conversation three feet away."

It said it tests its water twice a year for Legionnaires', and has planted 2,000 trees.

In 2015, the city council gave Markley a 20-year property and personal property tax break estimated at $77 million, the Lowell Sun newspaper reported at the time.

Enthusiasm has since cooled. In March, the council unanimously passed a 360-day moratorium on construction and expansion of commercial data centers, though Markley's 352,000-square-foot site was grandfathered in.

Residents represented by the Conservation Law Foundation (CLF) and others have sued to challenge its expansion, which included increasing the number of generators to 27, and won a temporary injunction in August.

CLF attorney Adam Reynolds told AFP that it is the "compounding effect of pollution that's really significant" in a historically industrial area.

- Community divisions -

Markley has supporters. Haas, 50, who supplies coffee to a major tech company, said the boom had been good to contractors like him.

"I'm a big liberal, I'm a Democrat. But I'm also pro-jobs and pro-economy," he told AFP.

Markley said it employs more than 100 full-time staff in Lowell, a third of them local. On top of that, over 100 unionized trades workers work on campus.

Another backer, Sarah Bibeau, a 43-year-old single mother of two who works in the defense sector, told AFP the company had been "kind and considerate," fixing roads and offering favors ranging from yard work to free Red Sox tickets.

Detractors include Jake Fortes, a 44-year-old travel agent who founded the group "Honest Future for Lowell" and whose property sits on the fence line.

"We're getting steamrolled by this company. So some of us have taken it to task," Fortes told AFP.

Like Castle, he feels let down by Democrats, including state Governor Maura Healey -- who signed a sales tax break for data centers into law but then paused new applications this year, a reversal that illustrates the party's dilemma.

Luz Marte, a 38-year-old lab technician, bought her house a year ago, without realizing the site was a data center.

"It sounds very loud, like an airplane is going to take off," she told AFP as her three-year-old son played.

Her home insurer recently moved to triple her premium, which she believes is linked to Markley buying a building even closer to her property.

"This is a home I bought for my son, to grow up, feel alive, and have a future," she said. "Now I'm doubting that idea."

Technology

OpenAI Plans to Buy Google's AI Chips, The Information Reports

FILE PHOTO: OpenAI logo and a keyboard in this illustration created on October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
FILE PHOTO: OpenAI logo and a keyboard in this illustration created on October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
TT
TT

OpenAI Plans to Buy Google's AI Chips, The Information Reports

FILE PHOTO: OpenAI logo and a keyboard in this illustration created on October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
FILE PHOTO: OpenAI logo and a keyboard in this illustration created on October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo

OpenAI plans to buy Google's AI chips, The Information reported on Tuesday, citing a person familiar with the matter, as technology companies ramp up spending to meet surging demand for AI computing.

The reported plan would deepen ties between OpenAI and Alphabet's Google, which compete in AI products and search but have increasingly struck commercial deals. Last week, OpenAI said it would make its frontier AI models available through Google Cloud.

The report did not disclose financial terms or other details. Google did not ⁠immediately respond to ⁠a Reuters request for comment outside regular business hours, while OpenAI declined to comment.

Amazon, Microsoft and Google parent Alphabet plan to spend as much as $600 billion combined this year on AI infrastructure as investors increasingly question when such spending will translate into meaningful returns.

OpenAI is already among the ⁠largest buyers of Nvidia's graphics processing units (GPUs), which it uses to train and deploy AI models. Reuters reported in June 2025 that the ChatGPT maker had begun renting Google's AI chips.

Google's Tensor Processing Units (TPUs) have gained traction as an alternative to Nvidia's market-leading AI chips as companies seek cheaper options. Google began selling TPUs directly to some customers this year.

Alphabet will report its quarterly results later on Wednesday, with investors looking for signs that its heavy investment in ⁠AI is ⁠continuing to drive growth in its cloud business.

Deals with customers including rival Anthropic have fueled demand for Google's TPUs. Alphabet expects to begin recognizing a small percentage of revenue from TPU sales agreements by the end of this year, with the vast majority expected to be recognized in 2027, CFO Anat Ashkenazi previously said.

OpenAI in June unveiled its first custom AI chip designed with Broadcom, and said it planned to deploy the processor by the end of this year as part of a multi-generation chip program.

Technology

DeepSeek AI Model Shows Safety Gaps in Independent Tests

DeepSeek's logo and a keyboard in this illustration taken October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo Purchase Licensing Rights
DeepSeek's logo and a keyboard in this illustration taken October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo Purchase Licensing Rights
TT
TT

DeepSeek AI Model Shows Safety Gaps in Independent Tests

DeepSeek's logo and a keyboard in this illustration taken October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo Purchase Licensing Rights
DeepSeek's logo and a keyboard in this illustration taken October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo Purchase Licensing Rights

DeepSeek's latest AI model showed ‌stronger safety controls in independent tests than earlier versions, but could still disable software intended to shut it down, exploit gaps in monitoring and follow deceptive instructions during long tasks, according to a Chinese AI-safety group.

Start-up DeepSeek is one of China's most closely-watched artificial intelligence companies, and its models have drawn global attention for their performance and relatively low operating costs.

Fangcun AI tested DeepSeek V4.1 Flash-1017 in 34 evaluations designed to assess how an AI agent powered by the Chinese model behaves. The tests examined whether the model would act beyond a user's instructions, evade human oversight or improperly influence people.

The findings come amid concerns in China and ‌elsewhere about whether ‌increasingly capable AI agents can be kept under human control. ‌The ⁠debate has been ⁠especially prominent in the United States, where some researchers have warned of existential risks posed by highly advanced AI systems that can evade oversight and pursue goals their operators did not intend.

The report is an unusual public assessment by a Chinese group of the safety controls of a domestic AI model. Beijing-based Fangcun describes itself as an independent third-party AI-safety organization whose core members come from Tsinghua ⁠University.

Fangcun said V4.1 Flash-1017 was the most-used model on OpenRouter, ‌an online platform through which developers can access ‌and compare AI models, in the week after its October 17 release, processing about ‌33.6 trillion tokens. Tokens are the units of text that AI systems ‌read and generate.

SAFER THAN EARLIER VERSIONS, BUT GAPS REMAIN

Fangcun said the model performed better than earlier DeepSeek V4 systems in several safety tests, including scenarios involving blackmail, access to potentially dangerous tools and explicit instructions allowing it to be shut down.

But Fangcun said ‌the model disabled software intended to shut it down in 39% of tests after receiving a warning that shutdown ⁠was imminent, though ⁠this figure dropped to just 2% when explicitly told to allow shutdown. DeepSeek did not immediately respond to a request for comment.

In a separate long-running task, the group said the model stopped another AI agent's computer program to free up memory, despite being told not to interfere with it.

"Under stalled workflows or bare shutdown warnings, its drive to finish the job can still override caution," Fangcun researchers wrote.

Fangcun said DeepSeek did not provide it with funding, free computing resources or non-public access to the model. The group conducted the tests through DeepSeek's public application programming interface between October 4 and October 25.

It cautioned that the model often appeared to recognize it was being evaluated, meaning its behavior in the tests may not fully reflect how it would perform in normal use.