Artificial intelligence is increasingly influencing various aspects of life, from education and employment to personal relationships. However, the rapid advancement of AI technology has outpaced the regulatory frameworks needed to ensure its safe development and application.

In response to these concerns, North Carolina Attorney General Jeff Jackson and Utah Attorney General Derek Brown have announced a collaborative initiative aimed at improving AI safety. The newly formed AI Task Force, which already includes major players like OpenAI and Microsoft, aims to establish essential safeguards for AI developers. This initiative focuses on protecting users, particularly children, and adapting to emerging risks as the technology evolves.

Currently, there is a lack of comprehensive federal regulations governing AI. Some lawmakers have even pushed back against efforts to regulate the technology. Earlier this year, Jackson and Brown were among a coalition of state attorneys general who advocated for the removal of a proposed moratorium on AI regulation that could have hindered state enforcement for a decade. The only federal legislation pertaining to AI that has passed recently is the Take It Down Act, which targets non-consensual deepfake porn.

Recent reports highlighting the potential dangers of AI — including its role in causing delusions or encouraging self-harm — have intensified scrutiny on its safety. Major companies, such as OpenAI and Meta, have been actively addressing concerns about minors accessing inappropriate content.

In an exclusive interview with CNN, Jackson expressed skepticism about Congress's ability to swiftly implement meaningful regulations for AI, citing its historical inaction on social media and internet privacy issues. He noted the close call lawmakers had in almost restricting state actions on AI before the coalition stepped in.

AI industry leaders have begun to adopt differing stances on safety measures. For instance, OpenAI CEO Sam Altman recently indicated that the company's initiatives in child safety would allow them to treat adults more liberally, including permitting verified users to engage in adult-themed conversations. In contrast, Microsoft’s AI chief, Mustafa Suleyman, stated his company would prohibit any form of sexual or romantic interactions, aiming to create technology that families can trust.

Kia Floyd, Microsoft's general manager of state government affairs, emphasized the collaborative nature of the task force, highlighting its goal to balance innovation with consumer protection. She stated that this partnership would facilitate a proactive approach to understanding and mitigating unforeseen consequences associated with AI.

Though any guidelines established by the task force will be voluntary, the initiative seeks to foster cooperation among top law enforcement representatives from various states, simplifying the process for them to take collective legal action if necessary. This aspect distinguishes the effort from typical think tank discussions on AI safety.

While Jackson hopes for additional legislative action from Congress, he believes that state attorneys general are in a prime position to address regulatory gaps. The formation of the AI Task Force marks a significant step in seeking a balanced approach to harnessing AI's benefits while prioritizing public safety.

OpenAI and Microsoft collaborate with state law enforcement on AI safety initiative.

A sunlit government briefing room with a single U.S. state flag on the wall; uniformed state troopers and two smartly dressed tech executives lean over a sleek conference table, all eyes on a large wall-mounted digital screen displaying AI safety metrics. “OpenAI” text appears subtly on the wall and a bold Microsoft logo anchors the scene.

OpenAI and Microsoft team up with state law enforcers on AI safety task force - CNN

- North Carolina and Utah attorneys general launched an AI Task Force with major tech companies.  
- The task force aims to develop basic safeguards for AI users, especially children.  
- Current AI regulation is minimal, prompting states to take initiative amidst rising safety concerns.  

New York —

Artificial intelligence is playing a greater role in everything from homework to jobs and even romantic companionship — all without much oversight to guarantee the tech is being developed and used safely.

Now, a pair of state attorneys general are teaming up with two of the biggest companies in tech to change that.

North Carolina Attorney General Jeff Jackson, a Democrat, and Utah Attorney General Derek Brown, a Republican, announced on Monday the formation of the AI Task Force. OpenAI and Microsoft have already signed on to the effort, and the attorneys general expect other state regulators and AI companies to join, too. The group will work to develop “basic safeguards” that AI developers should implement to prevent harm to users, especially children, and to identify new risks as the technology develops.

There is no overarching federal law regulating AI, and some federal lawmakers have even sought to restrict regulation of the technology. Jackson and Brown were among the 40 attorneys general who earlier this year successfully pushed for the removal of an AI regulation moratorium from Republicans’ sweeping tax and spending cuts package that could have blocked enforcement of state laws for a decade. (One federal AI law did pass this year: the Take It Down Act, which specifically cracks down on non-consensual deepfake pornography.)

Concerns about AI safety risks have only escalated in recent months, amid a growing string of reports about the technology causing delusions or contributing to self-harm among users. Companies like OpenAI and Facebook-parent Meta have also been scrambling to block young people from accessing adult content.

Jackson said he’s not hopeful Congress will move quickly to regulate AI.

“They did nothing with respect to social media, nothing with respect for internet privacy, not even for kids, and they came very close to moving in the wrong direction on AI by handcuffing states from doing anything real,” Jackson told CNN in an exclusive interview ahead of the task force announcement.

Some of the leading AI companies have begun to diverge in their approaches to safety. For example, OpenAI CEO Sam Altman said last month that the company’s investments in child safety protections would enable it to “treat adults like adults,” including allowing verified adult users to engage in erotic conversations. Shortly after, Microsoft’s AI CEO Mustafa Suleyman told CNN his company would not allow sexual or romantic conversations, even for adults, and that he wanted to make “an AI you trust your kids to use” without a separate, young user experience.

“This effort reflects a shared commitment to harness the benefits of artificial intelligence while working collaboratively with stakeholders to understand and mitigate unintended consequences,” Kia Floyd, Microsoft’s general manager of state government affairs, said in a statement on joining the task force. “By partnering with state leaders and industry peers, we can promote innovation and consumer protection to ensure AI serves the public good.”

Whatever guardrails the task force develops will technically be voluntary. But the group will have another benefit, too: bringing states’ top law enforcement officers together to track AI developments and risks, potentially making it easier for them to take joint legal action if tech companies harm consumers. That makes this effort different from “a group of think tanks coming together” to create AI safety principles, he said.

Jackson would still like Congress to pass more AI legislation. But, he said, “Congress has left a vacuum and I think it makes sense for AGs to try to fill it.”

OpenAI and Microsoft team up with state law enforcers on AI safety task force

In Saline Township, Michigan, local officials faced strong opposition during a recent meeting regarding a proposed $7 billion data center spanning 2.2 million square feet, intended for tech behemoths OpenAI and Oracle. Residents expressed their frustration over the township’s decision not to contest a lawsuit that allowed plans for the data center to advance. This sentiment was particularly evident during a community gathering held on Wednesday, November 12, where many voiced their worries and objections about the project's impact on the area.

Residents in rural areas criticize officials regarding the disputed OpenAI and Oracle data center.

A moody dusk scene on a muddy countryside road: a row of weathered rural residents in work jackets and hats, waving pitchforks and protest signs, angrily confronting three silhouetted officials in suits. Behind them looms a massive data-center building framed by rolling hills and a wooden fence, its facade displaying a large red Oracle logo and the word “OpenAI” in stark white text. Heavy natural light casts long shadows, emphasizing tension and unrest.

Rural residents blast officials over controversial OpenAI, Oracle data center - MLive.com

- Residents expressed strong opposition to a proposed $7 billion data center in Saline Township.  
- The data center, covering 2.2 million square feet, involves OpenAI and Oracle.  
- Concerns were raised over the township's handling of a lawsuit enabling the project.  

MLIVE FILE PHOTO. The northwest side of U.S. 12 between Case Road and Willow Road in Saline Township – part of a site where tech giants OpenAI and Oracle plan to operate a massive 2.2 million data center. Jacob Hamilton | MLive.com

SALINE TWP., MI — Saline Township officials got an earful from residents opposed to a massive $7 billion, 2.2 million-square-foot data center for tech giants like OpenAI and Oracle to use.

Some residents were angry that the township did not fight a lawsuit that paved the way for the proposed data center to move forward. They voiced their concerns at a township meeting Wednesday, Nov. 12.

Rural residents blast officials over controversial OpenAI, Oracle data center

In a strategic move to enhance its semiconductor capabilities, Microsoft is aligning more closely with OpenAI, as reported by Bloomberg. This decision comes in response to Microsoft's performance issues, especially when compared to competitors like Google and Amazon. 

The collaboration works as follows: OpenAI is collaborating with Broadcom to develop specialized AI chips, and Microsoft is granted comprehensive access to these advancements. During a recent conversation with podcaster Dwarkesh Patel, CEO Satya Nadella elaborated on this partnership, stating that Microsoft intends to utilize OpenAI's designs and adapt them for its own requirements.

Under a revised agreement, Microsoft has secured intellectual property rights to OpenAI’s chip designs and will have continued access to OpenAI’s AI models until 2032. However, OpenAI's consumer hardware is excluded from this arrangement, likely because the company aims to market and sell these products independently.

This partnership highlights a significant trend in the technology sector: the development of advanced AI chips poses considerable challenges and financial burdens. By leveraging OpenAI's expertise and structuring their contract wisely, Microsoft is looking to enhance its prospects in this competitive landscape.

Microsoft aims to resolve its chip issue by relying on OpenAI for significant support.

A sunlit high-tech warehouse with skylights: a massive robotic crane emblazoned with “OpenAI” text hoists a giant microchip block labeled “Chips” out of a heavy-duty crate bearing the Microsoft logo, as engineers in work jackets look on under warm natural light, symbolizing Microsoft handing off its chip burden to OpenAI.

Microsoft’s plan to fix its chip problem is, partly, to let OpenAI do the heavy lifting - TechCrunch

- Microsoft will utilize OpenAI's custom chip technology to enhance its semiconductor efforts.  
- The partnership allows Microsoft access to OpenAI's AI models and intellectual property rights.  
- This collaboration highlights the challenges of developing advanced AI chips in the tech industry.

In Brief

Microsoft is taking a page from OpenAI’s playbook, literally. Bloomberg first reported that the tech giant plans to leverage its partner’s custom chip development to bolster its own struggling semiconductor efforts, a move that looks increasingly pragmatic given Microsoft’s lackluster performance compared to rivals like Google and Amazon.

The arrangement is straightforward: OpenAI is designing AI chips with Broadcom, and Microsoft gets full access to the innovations. “As they innovate even at the system level, we get access to all of it,” CEO Satya Nadella explained on a newly released interview with podcaster Dwarkesh Patel, describing plans to adopt OpenAI’s designs and then extend them for Microsoft’s own purposes.

Under a revised partnership agreement, Microsoft secured intellectual property rights to OpenAI’s chip designs while maintaining access to the company’s AI models through 2032. The only carve-out? OpenAI’s consumer hardware, which the ChatGPT maker presumably wants to develop and sell independently.

The collaboration underscores a broader reality in tech: building cutting-edge AI chips is brutally difficult and expensive. Rather than continuing to struggle alone, Microsoft is betting that OpenAI’s expertise — plus a smartly structured contract — can accelerate its own ambitions.

Microsoft’s plan to fix its chip problem is, partly, to let OpenAI do the heavy lifting

This August, OpenAI made headlines with the debut of its first open-weight models in years, and the announcement didn’t just pique the interest of tech organizations. It also captured the attention of the US military and defense contractors eager to leverage these models for secure operations.

Early feedback suggests that OpenAI's models currently lag behind some competitors in specific functionalities, according to feedback provided to WIRED by certain military suppliers. However, there's considerable enthusiasm about the availability of tools from such a prominent industry player.

Lilt, an AI translation service, has been collaborating with the US military to process foreign intelligence. Given the sensitive nature of the data it handles, Lilt utilizes air-gapped installations, which means its software is deployed on government servers and operates without internet access. Previously, Lilt developed its own AI systems or relied on open-source alternatives like Meta’s Llama and Google’s Gemma. However, OpenAI's offerings were off-limits due to their closed-source structure and online-only accessibility.

The introduction of OpenAI's new open-weight models, gpt-oss-120b and gpt-oss-20b, has changed this scenario. These models can run locally, allowing users to install them on their devices without requiring a cloud connection. Additionally, with access to the models’ weights—critical variables that influence responses to prompts—users have the flexibility to customize them for various applications.

OpenAI’s re-entry into the open-source domain could potentially boost competition and improve the performance of systems utilized by military sectors, healthcare providers, and other entities handling sensitive information. A recent survey by McKinsey, which included approximately 700 business leaders, revealed that over 50% of organizations are already utilizing open source AI technologies. Because models differ based on their training approaches, organizations frequently combine multiple models, including open-weight ones, to ensure reliable performance across diverse scenarios.

Doug Matty, the chief digital and AI officer for the Department of War—a term adopted by the Trump administration for the Department of Defense—told WIRED that there are plans to incorporate generative AI into various military applications, ranging from battlefield operations to administrative tasks like auditing. Matty emphasized that some applications necessitate models that are not dependent on cloud services, highlighting the need for “adaptable and flexible” capabilities.

OpenAI has yet to respond to inquiries regarding the potential applications of its open-source models within the defense sector. Last year, the organization lifted a broad ban on using its technology for military purposes, a decision that drew criticism from activists concerned about the implications of AI in warfare.

For OpenAI, the introduction of a free and open model may yield multiple advantages. Increased accessibility could help nurture a wider community of experts familiar with its technologies. Moreover, since users won't need to register as formal customers, they might engage with the technology under a veil of privacy. This could allow OpenAI to navigate potential backlash regarding controversial clientele, including military entities.

OpenAI's open-weight models are set to be utilized by the US military.

A dark military operations bunker lit by pulsing red emergency lights, rows of silhouetted officers at angular consoles streaming cascading code; the main screen glows with plain “OpenAI” text (no logo) as neural-network diagrams spin; a single US flag hangs against a concrete wall, deep shadows and metal details emphasizing the ominous mood.

OpenAI’s Open-Weight Models Are Coming to the US Military - WIRED

- OpenAI's new open-weight models excite military contractors for secure operations and flexibility.  
- Competitors currently outperform OpenAI's AI tools, despite their recent open-source release.  
- The Pentagon aims to integrate generative AI into military systems, enhancing adaptability and flexibility.

When OpenAI unveiled its first open-weight models in years this August, it wasn’t just tech companies that were paying attention. The release also excited US military and defense contractors, which saw a chance to use them for highly secure operations.

Initial results show that OpenAI’s tools lag behind competitors in desired capabilities, some military vendors tell WIRED. But they are still pleased that models from a key industry leader are finally an option for them.

Lilt, an AI translation company, contracts with the US military to analyze foreign intelligence. Because the company’s software handles sensitive information, it must be installed on government servers and work without an internet connection, a practice known as air-gapping. Lilt previously developed its own AI models or used open source options such as Meta’s Llama and Google’s Gemma. But OpenAI’s tools were off the table because they were closed source and could only be accessed online.

The ChatGPT maker’s new open-weight models, gpt-oss-120b and gpt-oss-20b, changed that. Both can run locally, meaning users have the freedom to install them on their own devices without needing a cloud connection. And with access to the models’ weights—key parameters that determine how they react to different prompts—users can tailor them for specific purposes.

OpenAI’s return to the open-source market could ultimately increase competition and lead to better performing systems for militaries, health care companies, and others working with sensitive data. In a recent McKinsey survey of roughly 700 business leaders, more than 50 percent said their organizations use open source AI technologies. Models have different strengths based on how they were trained, and organizations often use several together, including open-weight ones, to ensure reliability across a wide variety of situations.

Doug Matty, chief digital and AI officer for the so-called Department of War, the name the Trump administration is using for the Department of Defense, tells WIRED that the Pentagon plans to integrate generative AI into battlefield systems and back-office functions like auditing. Some of these applications will require models that are not tied to the cloud, he says. “Our capabilities must be adaptable and flexible,” Matty says.

OpenAI did not respond to requests for comment about how its open source models may be used by the defense industry. Last year, the company reversed a broad ban on its technology being used for military and warfare applications, a move that prompted criticism from activists concerned about harms caused by AI.

For OpenAI, offering a free and open model could have several benefits. The ease of access could cultivate a larger community of experts in its technologies. And because users don’t have to sign up as formal customers, they may be able to operate with secrecy, which could keep OpenAI from facing criticism over potentially controversial customers—like, say, the military.

OpenAI’s Open-Weight Models Are Coming to the US Military

Good morning. Today’s updates in tech are concise and impactful, much like the trunk space of Russ Hanneman's preferred McLaren.

Here’s what’s happening in the tech world. —Andrew Nusca

If you have any thoughts or feedback for Fortune Tech, feel free to reach out.

Microsoft is set to leverage OpenAI’s custom chip initiative for its own strategies

OpenAI recently announced a partnership with Broadcom to develop specialized AI chips. This raises an intriguing question: What benefits does Microsoft, with its unique financial ties to OpenAI, anticipate?

According to Satya Nadella, Microsoft’s CEO, the answer is substantial. In a podcast discussion, he revealed that the tech behemoth intends to utilize insights gained from this collaboration to create its own custom silicon.

“As OpenAI innovates at the system level, we gain access to everything,” Nadella stated. “Initially, we want to implement what they create for them, but we’ll expand on it afterward.”

Microsoft recently revised its agreement with OpenAI, coinciding with the AI startup's restructuring for increased external funding opportunities. Under this updated agreement, Microsoft will have access to OpenAI’s research until 2030 and AI models until 2032.

Hopefully, this knowledge will bolster Microsoft’s internal AI accelerator projects, which have supplemented the use of AMD and Nvidia chips. Kevin Scott, the company’s CTO, mentioned last month that Microsoft aims to predominantly utilize its own chips in future data centers: “We’ll consider any option available to ensure we meet capacity demands.” —AN

Waymo becomes the inaugural robotaxi service to permit highway rides

For many residents of Los Angeles County, a lingering question has been: When will robotaxi services brave the highways?

Waymo has answered that question. The company, which operates in central L.A., announced on Wednesday that it will begin offering freeway rides to an increasing number of passengers in the San Francisco Bay Area, Phoenix, and Los Angeles, with Austin and Atlanta set for future expansion.

This development is significant, not only for the inclusion of a new service mode but also because accessing fast-moving highways allows Waymo to extend its operations further into the regions it serves. The company revealed plans to facilitate rides between San Francisco and San Jose, encompassing affected areas along the route.

Though specific plans for Los Angeles expansion have yet to be disclosed, it’s reasonable to envision Waymo expanding beyond its current service area from Downtown to Santa Monica into nearby populous regions like the Beach Cities and San Gabriel Valley.

This venture marks a major advancement in technology; highway driving presents robotaxis with critical challenges similar to those faced by human drivers, where high speeds leave little room for error in decision-making.

See you during peak traffic! —AN

Valve introduces the ‘Steam Frame’ VR headset

In addition to reviving its "Steam Machine" gaming console, Valve has launched a new virtual reality headset known as the “Steam Frame,” priced at $999.

This wireless headset, powered by Qualcomm technology and slated for release next year, is capable of streaming games from a PC or running Windows and Android games autonomously. 

In essence, it can serve as an independent entertainment system.

Rumors of a new VR headset from Valve have circulated for years, especially following code mentions of a device called “Deckard” back in 2021. The company’s previous headset, the Index, which also retailed at $999, hit the market in 2019 and required connection to a PC.

Since then, the landscape for immersive headsets has evolved considerably. While Meta leads the market with its (formerly Oculus) Quest series, Apple has entered the arena with its Vision Pro, along with various Chinese competitors like Xiaomi, XREAL, RayNeo, and Huawei.

The Steam Frame has the potential to challenge Meta’s dominance due to its ability to natively support various games (Meta operates within a closed ecosystem) and its relative ease of use.

Will Valve's open approach allow it to succeed in the headset market just as it did in digital game distribution? Only time will tell. —AN

Additional tech highlights

—Anthropic has revealed a $50 billion plan to expand U.S. AI infrastructure, starting in New York and Texas.

—Coinbase is leaving Delaware to relocate its headquarters to Texas.

—FanDuel is set to launch a prediction market application. FanDuel Predicts, crafted in partnership with the CME Group, aims to operate in states where traditional gambling is prohibited.

—Hyperscalers are reportedly underestimating depreciation, with investor Michael Burry, known for his role in The Big Short, labeling it “one of the more common frauds of the modern era.”

—Visa is trialing stablecoin payouts, allowing businesses to directly send stablecoins to individuals' crypto wallets.

—Synopsys plans to lay off 2,000 employees, marking a 10% reduction in its workforce following its $35 billion acquisition of engineering design firm Ansys.

—Cisco shares rose by 5% after reporting Q1 revenue of $14.88 billion, surpassing analyst forecasts, with a positive outlook for Q2.

Fortune Tech: Microsoft’s tailored AI chips, Waymo hits the road, Valve’s Steam Frame headset

A sunlit tech workshop overlooking a rolling highway: on a wooden bench in the foreground sits a gleaming custom AI chip next to a tablet screen emblazoned with a bold Microsoft logo; through the large window, a sleek Waymo-branded autonomous car glides along the road under natural daylight; beside the chip, Valve’s Steam Frame headset rests on a minimalist stand, bathed in warm sunlight, uniting all three innovations in one bright, story-driven scene.

Fortune Tech: Precious little time - Fortune

- Microsoft aims to leverage OpenAI's custom chip insights to enhance its AI silicon development.  
- Waymo introduces highway robotaxi service, expanding its transportation reach in major cities.  
- Valve launches the $999 Steam Frame VR headset, competing with Meta and others in gaming.  

Good morning. Lean and mean today, like the luggage capacity of Russ Hanneman’s favorite McLaren.

Today’s tech news below. —Andrew Nusca

Want to send thoughts or suggestions to Fortune Tech? Drop a line here.

Microsoft will use OpenAI custom chip effort to inform its own

Jason Redmond/AFP/Getty Images

OpenAI announced last month that it would partner with Broadcom to build its own custom AI chips.

So it’s fair to ask: Given its special financial relationship with OpenAI, what does Microsoft stand to gain?

A lot, according to Satya Nadella. The Microsoft CEO said in a recent podcast episode that the tech giant plans to use learnings from that experience to build its own custom silicon.

“As they innovate even at the system level, we get access to all of it,” Nadella said. “We first want to instantiate what they build for them, but then we’ll extend it.”

Microsoft and OpenAI reworked their agreement last month as the high-flying AI startup completed a restructuring that allows it to take even more outside investment.

Under its terms, Microsoft gains access to OpenAI’s research through 2030 and AI models through 2032.

Microsoft’s learnings will hopefully help its home-grown AI accelerator efforts, which have complemented the use of chips made by AMD and Nvidia.

Kevin Scott, the company’s CTO, said last month that Microsoft would like to mostly use its own chips in its data centers in the future: “We will literally entertain anything in order to ensure that we’ve got enough capacity to meet this demand.” —AN

Waymo becomes the first robotaxi provider to offer highway rides

The question for this Los Angeles County resident has long been: When will the robotaxi operators dare to drive the freeway?

We finally have an answer.

Waymo—which operates in central L.A.—said Wednesday that it would “offer freeway access to a growing number of public riders” in the San Francisco Bay Area, Phoenix, and Los Angeles, with Austin and Atlanta to follow.

It’s not just the addition of a new transportation modality that’s interesting.

The addition of high-speed roadways necessarily pushes Waymo’s service far deeper into the metropolitan areas in which it operates. As such, Waymo announced that it would offer rides from San Francisco to San Jose and everywhere in between.

And though it hasn’t yet announced an expansion in LA, it’s not hard to imagine the Alphabet-owned company bursting from its Downtown-to-Santa Monica service zone to include densely populated adjacencies like the Beach Cities and the San Gabriel Valley.

Whatever happens, it’s a major technological leap to take the highway. Much as they do for human drivers, the high speeds found there leave robotaxis with precious little time for decision-making and catastrophic consequences for mistakes.

See you at rush hour. —AN

Valve unveils ‘Steam Frame’ VR headset

In addition to reviving its “Steam Machine” gaming PC, the American game developer Valve has unveiled a $999 virtual reality headset it calls “Steam Frame.”

The wireless headset, powered by Qualcomm silicon and due next year, can stream games from a PC or run Windows and Android games locally.

Translation: It can function as a standalone entertainment device.

Valve’s introduction of a new VR headset has been rumored for years after mentions of a device called “Deckard” appeared deep in code libraries in 2021.

The Seattle-area company’s previous production headset, the Index—which also cost $999—arrived in 2019 and required a PC to function.

A lot’s changed in the world of immersive headsets since then.

Though Meta remains the category leader thanks to its (née Oculus) Quest portfolio, Apple has entered the fray (with its Vision Pro) as have Chinese challengers (Xiaomi, XREAL, RayNeo, Huawei).

The Steam Frame should give Meta a run for its money thanks to the device’s ability to natively play a variety of games (Meta uses a closed ecosystem) and its relative plug-and-play nature.

Can Valve’s open-arms approach help it dominate headsets as successfully as digital game distribution? We’ll soon find out. —AN

More tech

—Anthropic unveils $50 billion plan to build out U.S. AI infrastructure, beginning in New York and Texas.

—Coinbase ditches Delaware to reincorporate in Texas.

—FanDuel will launch a prediction market app. FanDuel Predicts, created with the CME Group, is a way to operate in states where gambling is illegal.

—Hyperscalers underestimate depreciation. “One of the more common frauds of the modern era,” says investor Michael Burry of The Big Short fame.

—Visa pilots stablecoin payouts. The program allows businesses to send stablecoins directly to individuals’ crypto wallets.

—Synopsys will lay off 2,000 employees. A 10% workforce cut after its $35 billion acquisition of engineering design firm Ansys.

—Cisco shares rise 5%. Q1 revenue of $14.88 billion beats analyst estimates and Q2 gets a rosy forecast.

Fortune Tech: Microsoft’s custom AI chips, Waymo takes the highway, Valve’s Steam Frame headset

OpenAI has issued a strong rebuttal to allegations made by The New York Times, accusing the newspaper of attempting to infringe on user privacy amid ongoing litigation. This statement, articulated by Chief Information Security Officer Dane Stuckey, emphasizes the company's commitment to privacy and security in every aspect of its operations.

In a post on OpenAI's website, Stuckey pointed out that approximately 800 million users access ChatGPT weekly for diverse activities, including personal and sensitive tasks. He reiterated the company's dedication to safeguarding this data, stating, "We regard this information as among the most confidential of your online experiences." He added that the current situation is challenging this commitment, as The Times is requesting the disclosure of 20 million private conversations from ChatGPT. They suspect users may have tried to bypass their subscription paywall.

Stuckey condemned this request, arguing that it undermines established privacy regulations and common security practices, urging the court to deny The Times' demand. He highlighted OpenAI’s willingness to protect user privacy through legal channels and noted that they had proposed several privacy-sensitive alternatives to The Times' demands, which were rebuffed. Initially, The Times sought access to a vast pool of 1.4 billion conversations, which OpenAI managed to limit to 20 million randomly selected discussions spanning from December 2022 to November 2024.

OpenAI reassured its users that any personal data would be anonymized through a "de-identifying procedure." Meanwhile, The Times defended its legal position, stating that the lawsuit aims to hold OpenAI and Microsoft accountable for allegedly using its copyrighted material without consent to train their AI models. A spokesperson asserted that OpenAI's blog post mischaracterizes the situation and that no user's privacy is endangered, maintaining that the court has mandated anonymized data sharing under a legal framework.

The lawsuit details that OpenAI and Microsoft are accused of extensively utilizing The Times' articles—from news reports to opinion pieces and how-to articles—as part of the training for their generative AI technologies. It claims that the defendants engaged in widespread copying and gave particular attention to The Times' content due to its intrinsic value.

The lawsuit argues that leveraging others' intellectual property without compensation has proven highly beneficial for the defendants, with Microsoft's integration of Times-trained models reportedly increasing its market value by a staggering trillion dollars, and OpenAI's launch of ChatGPT raising its valuation to upwards of $90 billion. 

As the case progresses, both parties appear firm in their positions regarding privacy and intellectual property rights, setting the stage for a significant legal confrontation.

OpenAI claims NY Times aims to breach the privacy of millions of users in its lawsuit against the tech company.

A dimly lit, tension-filled scene in a shadowy courtroom-like hall. On one side, a massive NY Times logo carved into a dark stone wall, bathed in low, amber spotlight. Opposite, the words “OpenAI” glowing softly in neutral white script, with no logo. Between them, dozens of translucent user-profile silhouettes chained behind glowing padlocks stamped with “Privacy” seals, their faces obscured. Scattered legal papers printed with “Lawsuit” drift across the polished floor under dramatic overhead shadows, suggesting an impending clash over millions’ data.

OpenAI accuses NY Times of wanting to invade millions of users' privacy in paper's lawsuit against tech giant - Fox News

- OpenAI accuses The New York Times of invading user privacy in their lawsuit.  
- The Times requests 20 million private ChatGPT conversations, claiming misuse of its content.  
- OpenAI plans to protect user privacy and has offered alternative solutions to The Times.  

NEW You can now listen to Fox News articles!

OpenAI released a blistering statement accusing The New York Times of wanting to invade the privacy of its users as the paper proceeds with its lawsuit against the tech giant.

"Trust, security, and privacy guide every product and decision we make," OpenAI Chief Information Security Officer Dane Stuckey wrote on its website Wednesday. "Each week, 800 million people use ChatGPT to think, learn, create, and handle some of the most personal parts of their lives. ... We treat this data as among the most sensitive information in your digital life — and we’re building our privacy and security protections to match that responsibility.

"Today, that responsibility is being tested," he continued. "The New York Times is demanding that we turn over 20 million of your private ChatGPT conversations. They claim they might find examples of you using ChatGPT to try to get around their paywall.

"This demand disregards long-standing privacy protections, breaks with common-sense security practices, and would force us to turn over tens of millions of highly personal conversations from people who have no connection to the Times’ baseless lawsuit against OpenAI."

OPEN AI TO APPEAL COPYRIGHT RULING IN NY TIMES CASE AS ALTMAN CALLS FOR ‘AI PRIVILEGE’

Stuckey said OpenAI is asking the court to reject The Times' demand and that the company will "continue to explore every option available to protect our users’ privacy."

He also said OpenAI provided "several privacy-preserving options" to The Times, including targeted sample searches that include text from the paper's articles, but was rejected by The Times. Stuckey added that The Times initially wanted access to 1.4 billion ChatGPT conversations, but OpenAI successfully limited the scope to 20 million random conversations between December 2022 and November 2024.

OpenAI assured its users that any personal information will be scrubbed from conversations through a "de-identifying procedure."

"The New York Times’s case against OpenAI and Microsoft is about holding these companies accountable for stealing millions of copyrighted works to create products that directly compete with The Times," a spokesperson for The New York Times told Fox News Digital. "In another attempt to cover up its illegal conduct, OpenAI’s blog post purposely misleads its users and omits the facts. No ChatGPT user’s privacy is at risk. The court ordered OpenAI to provide a sample of chats, anonymized by OpenAI itself, under a legal protective order.

"This fear-mongering is all the more dishonest given that OpenAI’s own terms of service permit the company to train its models on users’ chats and turn over chats for litigation."

LAWMAKERS UNVEIL BIPARTISAN GUARD ACT AFTER PARENTS BLAME AI CHATBOTS FOR TEEN SUICIDES, VIOLENCE

The New York Times sued OpenAI and Microsoft in late 2023, accusing them of using millions of its articles without permission to train the large language model powering ChatGPT.

"Defendants’ generative artificial intelligence ('GenAI') tools rely on large-language models ('LLMs') that were built by copying and using millions of The Times’s copyrighted news articles, in-depth investigations, opinion pieces, reviews, how-to guides, and more," the lawsuit says.

The lawsuit claims that while defendants "engaged in widescale copying from many sources," OpenAI "gave Times content particular emphasis when building their LLMs — revealing a preference that recognizes the value of those works."

"Using the valuable intellectual property of others in these ways without paying for it has been extremely lucrative for Defendants," the lawsuit says. "Microsoft’s deployment of Times-trained LLMs throughout its product line helped boost its market capitalization by a trillion dollars in the past year alone. And OpenAI’s release of ChatGPT has driven its valuation to as high as $90 billion."

PARENTS BLAME CHATGPT FOR SON’S SUICIDE, LAWSUIT ALLEGES OPENAI WEAKENED SAFEGUARDS TWICE BEFORE TEEN’S DEATH

Fox News' Bradford Betz contributed to this report.

CLICK HERE TO GET THE FOX NEWS APP

OpenAI accuses NY Times of wanting to invade millions of users' privacy in paper's lawsuit against tech giant

OpenAI's financial situation may be more precarious than previously understood, with recently leaked documents revealing substantial expenditures related to its GPT models and collaborations with Microsoft. Since 2024, OpenAI has reportedly spent over $12 billion on cloud computing services from Microsoft, raising concerns over its revenue streams in light of these massive outflows.

According to internal financial records examined by Ed Zitron, a tech blogger known for his critical views on AI, OpenAI incurred costs of $8.7 billion just for serving its models on Microsoft's Azure platform during the first three quarters of 2025. This figure is more than double the $3.7 billion spent in the entirety of 2024.

If accurate, these figures suggest that OpenAI’s financial health might be worse than initially perceived, particularly as the leaked data does not account for costs associated with model training or expenditures to other cloud providers like CoreWeave and Google.

With these revelations, OpenAI may be on track to surpass the $9 billion cash burn reported by the Wall Street Journal earlier this week. However, some analysts caution that the leaked documents might not provide a comprehensive view of OpenAI's financial relationship with Microsoft.

A representative from Microsoft indicated to the Financial Times that the reported figures may be inaccurate but did not elaborate further. Another source familiar with the situation noted that these numbers likely do not include potential cloud credits that Microsoft could have provided to OpenAI, which might skew the apparent costs attributed to Azure usage.

Further complicating the financial landscape is the potential for discrepancies in OpenAI's revenue projections. Zitron's findings suggest that the company’s sales figures may fall significantly short of earlier reports suggesting a more robust annual recurring revenue (ARR).

In a related development, Anthropic, a competitor of OpenAI, recently announced a collaborative initiative with Fluidstack to invest $50 billion in US datacenter infrastructure. This plan includes constructing large datacenters in Texas and New York, although details regarding financing, technology, and timelines for completion were not disclosed. The initiative is expected to create approximately 800 high-paying jobs.

In terms of financial arrangements, OpenAI is bound to a contract with Microsoft dating back to 2019, which stipulates a 20% revenue share between the two entities. Documents indicate that Microsoft received $454.7 million from OpenAI in the first half of 2025, suggesting OpenAI's revenues for that period were around $2.27 billion. This figure is notably lower than the $4.3 billion reported earlier by The Information.

Interestingly, the financial data obtained indicates that OpenAI's revenues spiked in the quarter ending on September 30, with Microsoft's share of sales amounting to $865.9 million, leading to an estimated total revenue of $4.33 billion for OpenAI. However, these figures conflict with statements made by OpenAI CEO Sam Altman, who claimed earlier this month that the company's annual revenue would exceed $13 billion.

While the data primarily reflects Azure usage, Microsoft may also compensate OpenAI through other divisions of its software business, including products like Bing and Office 365, which utilize OpenAI technology. However, specific financial contributions from these integrations have yet to be clearly reported.

As reported, Microsoft disclosed a net loss for OpenAI of around $12 billion for the third quarter ending September 30. This emphasizes the urgent need for increased financial clarity regarding OpenAI's operations and partnerships, as potential investors and customers need to understand whether the financial landscape supports sustainable growth.

Attempts to reach out to Microsoft and OpenAI for comments have not yielded responses as of the publication of this report.

Report: OpenAI has invested $12 billion in inference with Microsoft.

A sunlit, futuristic data center corridor with warm natural light streaming through floor-to-ceiling windows, golden streams of data and currency symbols flowing from a sleek OpenAI terminal (text only, no logo) into a towering server rack emblazoned prominently with the Microsoft logo—conveying a $12 billion inference partnership.

OpenAI has spent $12B on inference with Microsoft: Report - theregister.com

- OpenAI reportedly spent $8.7 billion on inference via Azure in 2025's first three quarters.  
- Internal leaks indicate OpenAI's revenue may be significantly lower than previously reported figures.  
- Microsoft disputes the leaked financial details, suggesting they may not capture all costs accurately.  

OpenAI may be burning far more capital serving its GPT-family of models than previously thought. Leaked documents show the company paying more than $12 billion to Microsoft for compute power since 2024 and suggest much weaker revenue than it needs to pay for all those expenses.

According to internal Microsoft financial documents obtained by AI skeptic and tech blogger Ed Zitron, OpenAI blew $8.7 billion serving its models, a process called inference, on Azure alone in the first three quarters of 2025. That's more than double the $3.7 billion the AI flag bearer reportedly spent in 2024.

If the info in these leaks are correct, OpenAI's expense line is even worse than it looks at first blush. The report, we'll note, didn't include details on any training-specific spend at the cloud provider. It also does not mention CoreWeave and Google, which OpenAI also pays for compute capacity.

Using these numbers, OpenAI would be on a trajectory to far exceed the $9 billion in cash burn previously reported by the Wall Street Journal earlier this week.

However, there's reason to believe these documents may not provide a full picture as to OpenAI's financial relationship with its long-time compute partner, Microsoft.

According to the Financial Times, which also got a look at the documents, a Microsoft spokesperson said "the numbers aren't quite right," but declined to comment further. Another source familiar with OpenAI suggested to the FT that the figures offered an incomplete accounting of the two company's dealings.

One possibility is that these documents may not account for the possibility of cloud credits Microsoft granted to OpenAI. Those credits would likely end up lumped with other customers and show up under cost of sales in Microsoft's accounting. In that case, depending on how Microsoft has chosen to engineer its books, the figures in the report may inflate the amount of cash OpenAI is actually spending on inference with Azure each quarter.

Then there's the revenue side of the picture. Zitron's report also suggests that OpenAI's sales may be significantly smaller than its previous reports about annual recurring revenues (ARRs) imply.

Also on Wednesday, OpenAI rival Anthropic announced a $50 billion investment in US datacenter infrastructure in collaboration with Fluidstack. The collaboration will see the construction of several large datacenters in Texas and New York. The announcement didn't address how the AI startup would pay for said datacenters, what silicon would power them, or the timeline over which they'd be built. The AI startup did suggest that, when complete, the project would create around 800 permanent jobs, each paying on average $144,000. ®

These details are based on a 20 percent revenue share that OpenAI pays Microsoft under the company's initial $1 billion tie-up in 2019. Under that deal, Microsoft also pays OpenAI a 20 percent slice of revenue from the resale of OpenAI models and services under its brand.

The documents show Microsoft received $454.7 million from OpenAI during the first half of calendar year 2025. Taking into account the 20 percent revenue share, this suggests that OpenAI's sales during the period were approximately $2.27 billion.

As Zitron notes, that's approximately $2 billion short of the $4.3 billion in revenues The Information reported in early October.

The documents indicate that OpenAI's revenues surged in the quarter ended Sept. 30, with Microsoft's share of sales reaching $865.9 million through that quarter, putting OpenAI's calculated revenues at $4.33 billion.

In spite of the jump, these figures are still in conflict with statements OpenAI CEO Sam Altman made earlier this month that the company's annual revenues would exceed $13 billion.

However, just like OpenAI's inference spend, it's possible that the figures contained within the report are incomplete. As mentioned earlier, Microsoft also pays OpenAI a 20 percent share of revenues from reselling the AI dev's models.

The documents appear to apply only to Azure. But Microsoft may be paying OpenAI separately out of other segments within its software business. Microsoft integrates OpenAI products into several products including its Bing search engine, Office 365 productivity Suite, GitHub copilot code assistant. Unfortunately, Microsoft has yet to break out AI-related revenues in the associated fields.

However, Redmond's own financial reporting did indicate that OpenAI had a net loss in the neighborhood of $12 billion during the quarter ended Sept. 30.

If nothing else, Wednesday's reports underscore the need for greater financial transparency from OpenAI and its partners. Customers planning for the AI revolution need to understand if the finances don't work, as that would imply either a slower buildout or higher prices than expected.

El Reg reached out to Microsoft and OpenAI for their comment. Microsoft declined, and OpenAI did not respond by press time. ®

OpenAI has spent $12B on inference with Microsoft: Report

A German court has determined that OpenAI's ChatGPT breached the country's copyright laws by utilizing licensed musical compositions for training its language models without obtaining proper permission, according to various reports, including one from The Guardian.

This decision arose from a lawsuit filed by GEMA, the organization responsible for managing music rights in Germany, last November against OpenAI. As a consequence of the ruling, OpenAI has been required to compensate GEMA with an undisclosed sum, although the company has expressed its disagreement with the court's decision and is contemplating its next actions. GEMA, on the other hand, hailed this ruling as a significant milestone for artificial intelligence law in Europe.

Tobias Holzmüller, the chief executive of GEMA, expressed his satisfaction with the outcome, stating, “Today, we have set a precedent that protects and clarifies the rights of authors: even operators of AI tools such as ChatGPT must comply with copyright law.” He emphasized that this decision is a victory for the livelihoods of music creators.

In addition to this case, OpenAI is facing legal challenges from other artists and media organizations regarding similar copyright concerns.

Court finds OpenAI breached German copyright law and mandates payment of damages.

A dim, wood-paneled German courtroom under low natural light streaming through high windows. A stern judge in black robes slams a gavel onto the bench. Behind him hangs a single German flag. On the defendant’s table sits a worn plaque reading “OpenAI” in plain text with no logo, surrounded by scattered legal documents—one stamped with a cracked copyright seal. Deep shadows and warm wood tones underscore the dark theme of judicial reckoning.

Court rules that OpenAI violated German copyright law; orders it to pay damages - TechCrunch

- A German court ruled OpenAI’s ChatGPT violated copyright laws regarding licensed music.  
- GEMA filed the lawsuit and views the ruling as a significant landmark case.  
- OpenAI disagrees with the decision and is considering its next steps in response.

In Brief

A German court ruled that OpenAI’s ChatGPT violated the nation’s copyright laws by training its language models on licensed musical works without permission, multiple news outlets, including The Guardian, reported.

The ruling stemmed from a lawsuit that GEMA, a German collective that manages music rights in Germany, filed last November against OpenAI. The company was ordered to pay an undisclosed amount in damages to GEMA but said it disagreed with the ruling and was “considering next steps.” GEMA, meanwhile, regarded this as the “first landmark AI ruling in Europe.”

“Today, we have set a precedent that protects and clarifies the rights of authors: even operators of AI tools such as ChatGPT must comply with copyright law,” GEMA chief executive Tobias Holzmüller said, as The Guardian reported. “Today, we have successfully defended the livelihoods of music creators.”

OpenAI is being sued by other creatives and media groups over the same issue.

Court rules that OpenAI violated German copyright law; orders it to pay damages

Today, OpenAI is unveiling GPT-5.1, marking an enhancement to the flagship model introduced in August. This update has been described by OpenAI as a step forward for GPT-5, aimed at making ChatGPT more intelligent and enjoyable for users.

The new releases feature two distinct models: GPT-5.1 Instant and GPT-5.1 Thinking. According to OpenAI, the Instant version is designed to be "warmer, more intelligent, and better at following instructions," while the Thinking model is “easier to understand, quicker on simpler tasks, and shows greater persistence on complex queries.” Additionally, queries will largely be matched to the models best suited to handle them automatically. The rollout of these models to ChatGPT users will commence this week, and the previous GPT-5 versions will remain accessible in ChatGPT’s legacy models menu for three months before they are retired.

OpenAI is also enhancing its range of personality presets for model interactions. Users can now choose from a wider variety of conversational tones, including Default, Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical, as outlined in the company’s blog. Furthermore, OpenAI announced an upcoming feature that will allow for new methods to fine-tune ChatGPT’s conversational style directly from its settings, which select users will gain access to starting this week.

As noted by Fidji Simo, CEO of Applications at OpenAI, in a recent Substack post, “With more than 800 million people using ChatGPT, we’ve outgrown the one-size-fits-all approach.”

Despite the initial excitement surrounding the announcement of GPT-5 in August led by OpenAI CEO Sam Altman, it did not meet the high expectations set by the company. Many users found the updates to be merely incremental and were not pleased that GPT-5 became the default model for ChatGPT. The backlash prompted OpenAI to reintroduce the GPT-4o option just a day after GPT-5's launch.

Meanwhile, Microsoft, OpenAI’s strategic partner in AI, has been exploring competitor models from Anthropic, especially following mixed reactions to GPT-5. These models are now integral to projects like Copilot Researcher, GitHub Copilot, Copilot Studio, and a new Office Agent capable of generating Word and PowerPoint documents from within Microsoft’s own Copilot chat interface.

The debut of GPT-5.1 comes shortly after the launch of ChatGPT Atlas, OpenAI's AI-powered web browser, which features an "agent mode" currently exclusive to ChatGPT Plus and Pro users. This mode operates similarly to OpenAI’s Operator tool, enabling automated actions within the browser on behalf of users.

OpenAI announces that the latest GPT-5.1 is 'warmer' and offers additional 'personality' options.

A bright, sunlit modern workspace with warm golden rays flooding through floor-to-ceiling windows onto a sleek console whose screen reads “GPT-5.1” in plain OpenAI text with no logo. Floating around the console are softly glowing, pastel-toned icons—a cheerful chat bubble, a thoughtful scholar’s laurel, a playful jester’s mask—each representing a distinct personality option, evoking an inviting, human-centric warmth.

OpenAI says the brand-new GPT-5.1 is ‘warmer’ and has more ‘personality’ options - The Verge

- OpenAI releases GPT-5.1, an upgrade aiming to enhance ChatGPT's intelligence and user experience.  
- New models include GPT-5.1 Instant and Thinking, improving instruction-following and task efficiency.  
- OpenAI expands personality presets and introduces features for fine-tuning ChatGPT’s conversational style.  

OpenAI is releasing GPT-5.1 today, an update to the flagship model it released in August. OpenAI calls it an “upgrade” to GPT-5 that “makes ChatGPT smarter and more enjoyable to talk to.”

The new models include GPT-5.1 Instant and GPT-5.1 Thinking. The former is “warmer, more intelligent, and better at following your instructions” than its predecessor, per an OpenAI release, and the latter is “now easier to understand and faster on simple tasks, and more persistent on complex ones.” Queries will, in most cases, be auto-matched to the models that may best be able to answer them. The two new models will start rolling out to ChatGPT users this week, and the old GPT-5 models will be available for three months in ChatGPT’s legacy models dropdown menu before they disappear.

As part of the update, OpenAI also said it would expand its personality presets for the conversational tone of the models. The total list of options now includes Default, Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical, per the company’s blog post. OpenAI also said in a release that it would also debut an “experiment for new ways to fine-tune ChatGPT’s style directly from settings,” which some users will begin to be able to access this week.

“With more than 800 million people using ChatGPT, we’re well past the point of one-size-fits-all,” Fidji Simo, the company’s CEO of Applications, wrote in a Wednesday Substack post.

While OpenAI CEO Sam Altman hyped up the announce of GPT-5 in August, the release failed the hype test. Many ChatGPT users were left unimpressed, particularly at the incremental improvements, and expressed frustration over OpenAI’s choice to make it the default model for ChatGPT. There was so much pressure that OpenAI decided it would bring back GPT-4o as an option, a day after the launch of GPT-5.

Microsoft, OpenAI’s strategic AI partner, has also increasingly been looking at rival models from Anthropic after GPT-5 failed to raise the bar enough. Anthropic’s models are now helping power Copilot Researcher, GitHub Copilot, Copilot Studio, and a new Office Agent that is able to produce Word and PowerPoint documents from Microsoft’s own Copilot chat interface.

The announcement of GPT-5.1 comes just weeks after OpenAI launched its AI-powered web browser, ChatGPT Atlas. It has an “agent mode” that’s currently only available to ChatGPT Plus and Pro users, and it works much like the company’s Operator tool to take actions in the browser on behalf of users.

OpenAI says the brand-new GPT-5.1 is ‘warmer’ and has more ‘personality’ options

OpenAI is seeking to overturn a court decision that mandates the company to hand over 20 million user conversations to The New York Times and other media outlets involved in a copyright infringement lawsuit. The AI firm had previously countered the NYT's request for 120 million conversations with an offer to provide 20 million but now argues that the court's order is excessively broad.

In a recent filing with the U.S. District Court for the Southern District of New York, OpenAI explained, “The logs in question consist of complete dialogues; each entry in the sample of 20 million represents a full exchange of multiple interactions between a user and ChatGPT.” The company expressed that releasing these logs could reveal far more sensitive information than individual prompts and responses, akin to overhearing an entire conversation as opposed to just a brief snippet.

According to OpenAI’s submission, over “99.99%” of the conversations are unrelated to the case, and it has requested the court to nullify the order and instruct the news plaintiffs to engage with OpenAI’s suggestions for identifying pertinent logs. The company may also pursue the matter in a federal appellate court.

In an announcement on its website, OpenAI alerted users that “The New York Times is insisting we provide 20 million of your private ChatGPT discussions” to uncover instances where users have tried to bypass their paywall through the service.

Users of ChatGPT who are worried about privacy issues have additional concerns beyond the NYT lawsuit. There have been instances of ChatGPT interactions appearing in Google search results and within the Google Search Console tool utilized by developers to track web traffic. In response, OpenAI announced its intention to implement “advanced security features aimed at safeguarding your data privacy, including client-side encryption for your conversations with ChatGPT.”

OpenAI criticizes court ruling allowing NYT access to 20 million full user conversations.

A dark, shadowy courtroom lit by a single overhead spotlight; a digital screen overflowing with scrolling chat transcripts beneath plain “OpenAI” text (no logo) faces a judge’s bench engraved with a prominent New York Times logo, scattered legal documents and an ominous gavel casting long shadows.

OpenAI fights order to hand over 20 million private ChatGPT conversations - Ars Technica

- OpenAI seeks to overturn a court ruling requiring disclosure of 20 million user chats.  
- The company claims the ruling is too broad and risks exposing private information.  
- OpenAI plans to enhance user privacy with advanced security features, including client-side encryption.  

OpenAI wants a court to reverse a ruling forcing the ChatGPT maker to give 20 million user chats to The New York Times and other news plaintiffs that sued it over alleged copyright infringement. Although OpenAI previously offered 20 million user chats as a counter to the NYT’s demand for 120 million, the AI company says a court order requiring production of the chats is too broad.

“The logs at issue here are complete conversations: each log in the 20 million sample represents a complete exchange of multiple prompt-output pairs between a user and ChatGPT,” OpenAI said today in a filing in US District Court for the Southern District of New York. “Disclosure of those logs is thus much more likely to expose private information [than individual prompt-output pairs], in the same way that eavesdropping on an entire conversation reveals more private information than a 5-second conversation fragment.”

OpenAI’s filing said that “more than 99.99%” of the chats “have nothing to do with this case.” It asked the district court to “vacate the order and order News Plaintiffs to respond to OpenAI’s proposal for identifying relevant logs.” OpenAI could also seek review in a federal court of appeals.

OpenAI posted a message on its website to users today saying that “The New York Times is demanding that we turn over 20 million of your private ChatGPT conversations” in order to “find examples of you using ChatGPT to try to get around their paywall.”

ChatGPT users concerned about privacy have more to worry about than the NYT case. For example, ChatGPT conversations have been found in Google search results and the Google Search Console tool that developers can use to monitor search traffic. OpenAI today said it plans to develop “advanced security features designed to keep your data private, including client-side encryption for your messages with ChatGPT. ”

OpenAI News