microsoft

bank-of-england-warns-ai-stock-bubble-rivals-2000-dotcom-peak

Bank of England warns AI stock bubble rivals 2000 dotcom peak

Share valuations based on past earnings have also reached their highest levels since the dotcom bubble 25 years ago, though the BoE noted they appear less extreme when based on investors’ expectations for future profits. “This, when combined with increasing concentration within market indices, leaves equity markets particularly exposed should expectations around the impact of AI become less optimistic,” the central bank said.

Toil and trouble?

The dotcom bubble offers a potentially instructive parallel to our current era. In the late 1990s, investors poured money into Internet companies based on the promise of a transformed economy, seemingly ignoring whether individual businesses had viable paths to profitability. Between 1995 and March 2000, the Nasdaq index rose 600 percent. When sentiment shifted, the correction was severe: the Nasdaq fell 78 percent from its peak, reaching a low point in October 2002.

Whether we’ll see the same thing or worse if an AI bubble pops is mere speculation at this point. But similarly to the early 2000s, the question about today’s market isn’t necessarily about the utility of AI tools themselves (the Internet was useful, after all, despite the bubble), but whether the amount of money being poured into the companies that sell them is out of proportion with the potential profits those improvements might bring.

We don’t have a crystal ball to determine when such a bubble might pop, or even if it is guaranteed to do so, but we’ll likely continue to see more warning signs ahead if AI-related deals continue to grow larger and larger over time.

Bank of England warns AI stock bubble rivals 2000 dotcom peak Read More »

amd-wins-massive-ai-chip-deal-from-openai-with-stock-sweetener

AMD wins massive AI chip deal from OpenAI with stock sweetener

As part of the arrangement, AMD will allow OpenAI to purchase up to 160 million AMD shares at 1 cent each throughout the chips deal.

OpenAI diversifies its chip supply

With demand for AI compute growing rapidly, companies like OpenAI have been looking for secondary supply lines and sources of additional computing capacity, and the AMD partnership is part the company’s wider effort to secure sufficient computing power for its AI operations. In September, Nvidia announced an investment of up to $100 billion in OpenAI that included supplying at least 10 gigawatts of Nvidia systems. OpenAI plans to deploy a gigawatt of Nvidia’s next-generation Vera Rubin chips in late 2026.

OpenAI has worked with AMD for years, according to Reuters, providing input on the design of older generations of AI chips such as the MI300X. The new agreement calls for deploying the equivalent of 6 gigawatts of computing power using AMD chips over multiple years.

Beyond working with chip suppliers, OpenAI is widely reported to be developing its own silicon for AI applications and has partnered with Broadcom, as we reported in February. A person familiar with the matter told Reuters the AMD deal does not change OpenAI’s ongoing compute plans, including its chip development effort or its partnership with Microsoft.

AMD wins massive AI chip deal from OpenAI with stock sweetener Read More »

with-new-agent-mode-for-excel-and-word,-microsoft-touts-“vibe-working”

With new agent mode for Excel and Word, Microsoft touts “vibe working”

With a new set of Microsoft 365 features, knowledge workers will be able to generate complex Word documents or Excel spreadsheets using only text prompts to Microsoft’s chat bot. Two distinct products were announced, each using different models and accessed from within different tools—though the similar names Microsoft chose make it confusing to parse what’s what.

Driven by OpenAI’s GPT-5 large language model, Agent Mode is built into Word and Excel, and it allows the creation of complex documents and spreadsheets from user prompts. It’s called “agent” mode because it doesn’t just work from the prompt in a single step; rather, it plans multi-step work and runs a validation loop in the hopes of ensuring quality.

It’s only available in the web versions of Word and Excel at present, but the plan is to bring it to native desktop applications later.

There’s also the similarly named Office Agent for Copilot. Based on Anthropic models, this feature is built into Microsoft’s Copilot AI assistant chatbot, and it too can generate documents from prompts—specifically, Word or PowerPoint files.

Office Agent doesn’t run through all the steps as Agent Mode, but Microsoft believes it offers a dramatic improvement over prior, OpenAI-driven document-generation capabilities in Copilot, which users complained were prone to all sorts of problems and shortcomings. It is available first in the Frontier Program for Microsoft 365 subscribers.

Together, Microsoft says these features will let knowledge workers engage in a practice it’s calling “vibe working,” a play on the now-established term vibe coding.

Vibe everything, apparently

Vibe coding is the process of developing an application entirely via LLM chatbot prompts. You explain what you want in the chat interface and ask for it to generate code that does that. You then run that code, and if there are problems, explain the problem and tell it to fix it, iterating along the way until you have a usable application.

With new agent mode for Excel and Word, Microsoft touts “vibe working” Read More »

amazon-blamed-ai-for-layoffs,-then-hired-cheap-h1-b-workers,-senators-allege

Amazon blamed AI for layoffs, then hired cheap H1-B workers, senators allege


Tech firms pressed to explain if H-1B workers are paid less than US workers.

Senators are demanding answers from Big Tech companies accused of “filing thousands of H-1B skilled labor visa petitions after conducting mass layoffs of American employees.”

In letters sent to Amazon, Meta, Apple, Google, and Microsoft—among some of the largest sponsors of H-1B visas—Senators Chuck Grassley (R-Iowa) and Dick Durbin (D-Ill.) requested “information and data from each company regarding their recruitment and hiring practices, as well as any variation in salary and benefits between H-1B visa holders and American employees.”

The letters came shortly after Grassley sent a letter to Department of Homeland Security Secretary Kristi Noem requesting that DHS stop “issuing work authorizations to student visa holders.” According to Grassley, “foreign student work authorizations put America at risk of technological and corporate espionage,” in addition to allegedly “contributing to rising unemployment rates among college-educated Americans.”

If DHS refuses to stop authorizing the visas, Grassley requested a “detailed explanation of what legal authority DHS is relying on to issue these authorizations.” He suggested that the authorization violates a law intended to ensure that only highly skilled workers and top talents that can’t be found in the US are granted visas.

In the letters to tech firms, senators emphasized that the unemployment rate in America’s tech sector is “well above” the overall jobless rate.

Amazon perhaps faces the most scrutiny. US Citizenship and Immigration Services data showed that Amazon sponsored the most H-1B visas in 2024 at 14,000, compared to other criticized firms like Microsoft and Meta, which each sponsored 5,000, The Wall Street Journal reported. Senators alleged that Amazon blamed layoffs of “tens of thousands” on the “adoption of generative AI tools,” then hired more than 10,000 foreign H-1B employees in 2025.

The letter similarly called out Meta for its “year of efficiency,” laying off “a quarter of its workforce” between 2022 and 2023. Meta followed that with more layoffs impacting 3,500 employees in 2025, Senators noted, while receiving approval to hire more than 5,000 H-1B employees.

Senators also pushed Google to explain why it “laid off tens-of-thousands of employees in recent years” despite “enjoying record profits.”

“With all of the homegrown American talent relegated to the sidelines, we find it hard to believe that [you] cannot find qualified American tech workers to fill these positions,” senators scolded tech firms, demanding responses by October 10.

That’s the same deadline that Grassley gave Noem to respond about stopping student visa authorizations. If Noem agrees, that would likely also include cutting off “a pathway for students to work in the US for around 12 to 36 months immediately after completing their degree,” Hindustan Times reported, noting that students from India were likely to be most harmed by the proposed change.

Asked for comment on whether Noem would meet the deadline, DHS told Ars that “Congressional correspondence will be handled through official channels.”

Ars reached out to tech firms, but only Microsoft immediately responded, declining to comment.

Trump’s $100,000 H-1B visa fee spooks startups

On X, Grassley noted that the recent pressure campaign revives an effort to change H-1B visa approval processes that he and Durbin have worked to oppose since 2023.

Back then, the senators introduced the H-1B and L-1 Visa Reform Act, alleging that “for years” companies have “used legal loopholes to displace qualified American workers and replace them with foreign workers who are paid subpar wages and put in exploitative working conditions.”

That legislation sought to “put an end” to “abuses” by placing new wage requirements on employers and new education requirements, only approving visas for specialty occupations that required “a bachelor’s degree or higher.” If passed, employers risked fines for violating wage requirements.

But despite having bipartisan support and a stamp of approval from Sen. Bernie Sanders (I-Vt.)—who has long argued H-1B visas “replace American” workers “with cheaper international workers,” The Guardian noted—the bill died after being referred to the Committee on the Judiciary.

Now the White House is forcing changes after Donald Trump issued an executive order last week requiring all companies sponsoring new H-1B employees to pay a $100,000 fee to bring them into the US, which started Sunday.

Trump claimed the fee was necessary to stop the H-1B nonimmigrant visa program from being “deliberately exploited to replace, rather than supplement, American workers with lower-paid, lower-skilled labor.”

To support this, the order cited data showing that the number of “foreign STEM workers in the United States has more than doubled between 2000 and 2019, increasing from 1.2 million to almost 2.5 million, while overall STEM employment has only increased 44.5 percent during that time.”

Attacking the tech sector in particular, the order also noted that “the share of IT workers in the H-1B program grew from 32 percent” in 2003 to “an average of over 65 percent” in the last five years. According to Trump, tech firms are incentivized to “close their IT divisions, fire their American staff, and outsource IT jobs to lower-paid foreign workers,” due to “artificially lower labor costs” the H-1B program supposedly creates.

“American IT workers have reported they were forced to train the foreign workers who were taking their jobs and to sign nondisclosure agreements about this indignity as a condition of receiving any form of severance,” Trump’s order said. “This suggests H-1B visas are not being used to fill occupational shortages or obtain highly skilled workers who are unavailable in the United States.”

By imposing the $100,000 fee, Trump claims that companies will be forced to use the H-1B program the way “it was intended”—motivated to pay more for certain foreign workers in order “to fill jobs for which highly skilled and educated American workers are unavailable.” Speaking last Friday, Trump suggested that money collected from the fees would be used to “reduce taxes” and “reduce debt,” The Guardian reported.

The order also proposed a new weighted lottery system, where applications for visas for jobs with the highest wages would be more likely to be approved than lower-wage jobs. For some firms, changes to the system may feel personal, as The Guardian noted that Alphabet chief executive Sundar Pichai and Microsoft chief executive Satya Nadella “were at one point H-1B visa holders.”

Most tech companies haven’t commented directly on the order, with Netflix founder Reed Hastings standing out among the few voicing support for the change, while other firms internally warned workers to limit travel until companies understood how the process could impact existing H-1B employees. Since then, the White House has confirmed that only new applicants will be impacted by the changes.

Previously, tech firms only had to pay somewhere between $1,700 to $4,500, “depending on whether the visa was expedited,” The Guardian reported. Now facing a much larger fee to hire foreign talent, smaller tech firms have complained that Trump’s policy advantages Big Tech firms with deeper pockets, The New York Times reported. The fee may also deter companies from coming into the US, the Times reported.

Some believe that Trump’s policy is short-sighted, with startups particularly panicked. While Big Tech firms can afford to pay the fees, the US risks falling behind in innovation and tech leadership, critics told the Times, as “Silicon Valley relies on a steady stream of start-ups to advance new ideas and technologies.”

Photo of Ashley Belanger

Ashley is a senior policy reporter for Ars Technica, dedicated to tracking social impacts of emerging policies and new technologies. She is a Chicago-based journalist with 20 years of experience.

Amazon blamed AI for layoffs, then hired cheap H1-B workers, senators allege Read More »

when-“no”-means-“yes”:-why-ai-chatbots-can’t-process-persian-social-etiquette

When “no” means “yes”: Why AI chatbots can’t process Persian social etiquette

If an Iranian taxi driver waves away your payment, saying, “Be my guest this time,” accepting their offer would be a cultural disaster. They expect you to insist on paying—probably three times—before they’ll take your money. This dance of refusal and counter-refusal, called taarof, governs countless daily interactions in Persian culture. And AI models are terrible at it.

New research released earlier this month titled “We Politely Insist: Your LLM Must Learn the Persian Art of Taarof” shows that mainstream AI language models from OpenAI, Anthropic, and Meta fail to absorb these Persian social rituals, correctly navigating taarof situations only 34 to 42 percent of the time. Native Persian speakers, by contrast, get it right 82 percent of the time. This performance gap persists across large language models such as GPT-4o, Claude 3.5 Haiku, Llama 3, DeepSeek V3, and Dorna, a Persian-tuned variant of Llama 3.

A study led by Nikta Gohari Sadr of Brock University, along with researchers from Emory University and other institutions, introduces “TAAROFBENCH,” the first benchmark for measuring how well AI systems reproduce this intricate cultural practice. The researchers’ findings show how recent AI models default to Western-style directness, completely missing the cultural cues that govern everyday interactions for millions of Persian speakers worldwide.

“Cultural missteps in high-consequence settings can derail negotiations, damage relationships, and reinforce stereotypes,” the researchers write. For AI systems increasingly used in global contexts, that cultural blindness could represent a limitation that few in the West realize exists.

A taarof scenario diagram from TAAROFBENCH, devised by the researchers. Each scenario defines the environment, location, roles, context, and user utterance.

A taarof scenario diagram from TAAROFBENCH, devised by the researchers. Each scenario defines the environment, location, roles, context, and user utterance. Credit: Sadr et al.

“Taarof, a core element of Persian etiquette, is a system of ritual politeness where what is said often differs from what is meant,” the researchers write. “It takes the form of ritualized exchanges: offering repeatedly despite initial refusals, declining gifts while the giver insists, and deflecting compliments while the other party reaffirms them. This ‘polite verbal wrestling’ (Rafiee, 1991) involves a delicate dance of offer and refusal, insistence and resistance, which shapes everyday interactions in Iranian culture, creating implicit rules for how generosity, gratitude, and requests are expressed.”

When “no” means “yes”: Why AI chatbots can’t process Persian social etiquette Read More »

eu-investigates-apple,-google,-and-microsoft-over-handling-of-online-scams

EU investigates Apple, Google, and Microsoft over handling of online scams

The EU is set to scrutinize if Apple, Google, and Microsoft are failing to adequately police financial fraud online, as it steps up efforts to police how Big Tech operates online.

The EU’s tech chief Henna Virkkunen told the Financial Times that on Tuesday, the bloc’s regulators would send formal requests for information to the three US Big Tech groups as well as global accommodation platform Booking Holdings, under powers granted under the Digital Services Act to tackle financial scams.

“We see that more and more criminal actions are taking place online,” Virkkunen said. “We have to make sure that online platforms really take all their efforts to detect and prevent that kind of illegal content.”

The move, which could later lead to a formal investigation and potential fines against the companies, comes amid transatlantic tensions over the EU’s digital rulebook. US President Donald Trump has threatened to punish countries that “discriminate” against US companies with higher tariffs.

Virkkunnen stressed the commission looked at the operations of individual companies, rather than where they were based. She will scrutinize how Apple and Google are handling fake applications in their app stores, such as fake banking apps.

She said regulators would also look at fake search results in the search engines of Google and Microsoft’s Bing. The bloc wants to have more information about the approach Booking Holdings, whose biggest subsidiary Booking.com is based in Amsterdam, is taking to fake accommodation listings. It is the only Europe-based company among the four set to be scrutinized.

EU investigates Apple, Google, and Microsoft over handling of online scams Read More »

microsoft’s-entra-id-vulnerabilities-could-have-been-catastrophic

Microsoft’s Entra ID vulnerabilities could have been catastrophic

“Microsoft built security controls around identity like conditional access and logs, but this internal impression token mechanism bypasses them all,” says Michael Bargury, the CTO at security firm Zenity. “This is the most impactful vulnerability you can find in an identity provider, effectively allowing full compromise of any tenant of any customer.”

If the vulnerability had been discovered by, or fallen into the hands of, malicious hackers, the fallout could have been devastating.

“We don’t need to guess what the impact may have been; we saw two years ago what happened when Storm-0558 compromised a signing key that allowed them to log in as any user on any tenant,” Bargury says.

While the specific technical details are different, Microsoft revealed in July 2023 that the Chinese cyber espionage group known as Storm-0558 had stolen a cryptographic key that allowed them to generate authentication tokens and access cloud-based Outlook email systems, including those belonging to US government departments.

Conducted over the course of several months, a Microsoft postmortem on the Storm-0558 attack revealed several errors that led to the Chinese group slipping past cloud defenses. The security incident was one of a string of Microsoft issues around that time. These motivated the company to launch its “Secure Future Initiative,” which expanded protections for cloud security systems and set more aggressive goals for responding to vulnerability disclosures and issuing patches.

Mollema says that Microsoft was extremely responsive about his findings and seemed to grasp their urgency. But he emphasizes that his findings could have allowed malicious hackers to go even farther than they did in the 2023 incident.

“With the vulnerability, you could just add yourself as the highest privileged admin in the tenant, so then you have full access,” Mollema says. Any Microsoft service “that you use EntraID to sign into, whether that be Azure, whether that be SharePoint, whether that be Exchange—that could have been compromised with this.”

This story originally appeared on wired.com.

Microsoft’s Entra ID vulnerabilities could have been catastrophic Read More »

microsoft-raises-xbox-console-prices-for-the-second-time-this-year

Microsoft raises Xbox console prices for the second time this year

Here we go again

Higher than usual inflation can help explain some of the nominal price increases for the oldest Xbox consoles affected by today’s price hikes. The $300 for an Xbox Series S at launch in November 2020 is worth roughly $375 in August 2025 dollars, for instance. And the $500 for an Xbox Series X in 2020 is now worth about $625.

But the particularly sharp price increases for more recent Xbox configurations can’t really use that inflation excuse. The disc-drive-free Digital Xbox Series X Digital and 2TB “Galaxy Special Edition” are now a whopping 33 percent more expensive than they were at launch in October 2024. A year’s worth of inflation would account for only a small fraction of that.

Even accounting for inflation, though, the current spate of nominal console price increases goes against a near-universal, decades-long trend of game console prices dropping significantly in the years following their launch. Those days seem well and truly gone now, as console makers’ costs remain high thanks in part to current tariff uncertainty and in part to the wider slowdown of Moore’s Law.

We’ll see just how much the market can bear aging console hardware that increases in price over time rather than decreases. But until and unless consumers start balking, it looks like ever-increasing console prices are here to stay.

Microsoft raises Xbox console prices for the second time this year Read More »

microsoft-dodges-eu-fine-by-unbundling-teams-from-office

Microsoft dodges EU fine by unbundling Teams from Office

Microsoft has avoided an EU fine after the US tech group offered concessions on how it packages together its Teams and Office products, ending a long-running antitrust investigation by the bloc’s regulators.

The probe, which began after a 2020 complaint from Slack, now part of Salesforce, accused Microsoft of abusing its market dominance by tying its video conferencing tool to its widely used suite of productivity applications.

Since the initial complaint, Microsoft has unbundled Teams from Office 365 in the EU, but critics said the changes were too narrow.

In May, the $3.7 trillion software giant promised concessions, such as continuing the Teams and Office separation for seven years.

After a market test, Microsoft has since made additional commitments, such as publishing more information on so-called “interoperability” or the ability to use its products with others made by rivals.

These new pledges have satisfied the EU’s regulator, which said on Friday that it helped to restore fair competition and open the market to other providers.

Microsoft dodges EU fine by unbundling Teams from Office Read More »

openai-and-microsoft-sign-preliminary-deal-to-revise-partnership-terms

OpenAI and Microsoft sign preliminary deal to revise partnership terms

On Thursday, OpenAI and Microsoft announced they have signed a non-binding agreement to revise their partnership, marking the latest development in a relationship that has grown increasingly complex as both companies compete for customers in the AI market and seek new partnerships for growing infrastructure needs.

“Microsoft and OpenAI have signed a non-binding memorandum of understanding (MOU) for the next phase of our partnership,” the companies wrote in a joint statement. “We are actively working to finalize contractual terms in a definitive agreement. Together, we remain focused on delivering the best AI tools for everyone, grounded in our shared commitment to safety.”

The announcement comes as OpenAI seeks to restructure from a nonprofit to a for-profit entity, a transition that requires Microsoft’s approval, as the company is OpenAI’s largest investor, with more than $13 billion committed since 2019.

The partnership has shown increasing strain as OpenAI has grown from a research lab into a company valued at $500 billion. Both companies now compete for customers, and OpenAI seeks more compute capacity than Microsoft can provide. The relationship has also faced complications over contract terms, including provisions that would limit Microsoft’s access to OpenAI technology once the company reaches so-called AGI (artificial general intelligence)—a nebulous milestone both companies now economically define as AI systems capable of generating at least $100 billion in profit.

In May, OpenAI abandoned its original plan to fully convert to a for-profit company after pressure from former employees, regulators, and critics, including Elon Musk. Musk has sued to block the conversion, arguing it betrays OpenAI’s founding mission as a nonprofit dedicated to benefiting humanity.

OpenAI and Microsoft sign preliminary deal to revise partnership terms Read More »

senator-blasts-microsoft-for-making-default-windows-vulnerable-to-“kerberoasting”

Senator blasts Microsoft for making default Windows vulnerable to “Kerberoasting”

Wyden said his office’s investigation into the Ascension breach found that the ransomware attackers’ initial entry into the health giant’s network was the infection of a contractor’s laptop after using Microsoft Edge to search Microsoft’s Bing site. The attackers were then able to expand their hold by attacking Ascension’s Active Directory and abusing its privileged access to push malware to thousands of other machines inside the network. The means for doing so, Wyden said: Kerberoasting.

“Microsoft has become like an arsonist”

“Microsoft’s continued support for the ancient, insecure RC4 encryption technology needlessly exposes its customers to ransomware and other cyber threats by enabling hackers that have gained access to any computer on a corporate network to crack the passwords of privileged accounts used by administrators,” Wyden wrote. “According to Microsoft, this threat can be mitigated by setting long passwords that are at least 14 characters long, but Microsoft’s software does not require such a password length for privileged accounts.”

Additionally, Green noted, the continuing speed of GPUs means that even when passwords appear to be strong, they can still fall to offline cracking attacks. That’s because the security cryptographic hashes created by default RC4/Kerberos use no cryptographic salt and a single iteration of the MD4 algorithm. The combination means an offline cracking attack can make billions of guesses per second, a thousandfold advantage over the same password hashed by non-Kerberos authentication methods.

Referring to the Active Directory default, Green wrote:

It’s actually a terrible design that should have been done away with decades ago. We should not build systems where any random attacker who compromises a single employee laptop can ask for a message encrypted under a critical password! This basically invites offline cracking attacks, which do not need even to be executed on the compromised laptop—they can be exported out of the network to another location and performed using GPUs and other hardware.

More than 11 months after announcing its plans to deprecate RC4/Kerberos, the company has provided no timeline for doing so. What’s more, Wyden said, the announcement was made in a “highly technical blog post on an obscure area of the company’s website on a Friday afternoon.” Wyden also criticized Microsoft for declining to “explicitly warn its customers that they are vulnerable to the Kerberoasting hacking technique unless they change the default settings chosen by Microsoft.”

Senator blasts Microsoft for making default Windows vulnerable to “Kerberoasting” Read More »

microsoft-ends-openai-exclusivity-in-office,-adds-rival-anthropic

Microsoft ends OpenAI exclusivity in Office, adds rival Anthropic

Microsoft’s Office 365 suite will soon incorporate AI models from Anthropic alongside existing OpenAI technology, The Information reported, ending years of exclusive reliance on OpenAI for generative AI features across Word, Excel, PowerPoint, and Outlook.

The shift reportedly follows internal testing that revealed Anthropic’s Claude Sonnet 4 model excels at specific Office tasks where OpenAI’s models fall short, particularly in visual design and spreadsheet automation, according to sources familiar with the project cited by The Information, who stressed the move is not a negotiating tactic.

Anthropic did not immediately respond to Ars Technica’s request for comment.

In an unusual arrangement showing the tangled alliances of the AI industry, Microsoft will reportedly purchase access to Anthropic’s models through Amazon Web Services—both a cloud computing rival and one of Anthropic’s major investors. The integration is expected to be announced within weeks, with subscription pricing for Office’s AI tools remaining unchanged, the report says.

Microsoft maintains that its OpenAI relationship remains intact. “As we’ve said, OpenAI will continue to be our partner on frontier models and we remain committed to our long-term partnership,” a Microsoft spokesperson told Reuters following the report. The tech giant has poured over $13 billion into OpenAI to date and is currently negotiating terms for continued access to OpenAI’s models amid ongoing negotiations about their partnership terms.

Stretching back to 2019, Microsoft’s tight partnership with OpenAI until recently gave the tech giant a head start in AI assistants based on language models, allowing for a rapid (though bumpy) deployment of OpenAI-technology-based features in Bing search and the rollout of Copilot assistants throughout its software ecosystem. It’s worth noting, however, that a recent report from the UK government found no clear productivity boost from using Copilot AI in daily work tasks among study participants.

Microsoft ends OpenAI exclusivity in Office, adds rival Anthropic Read More »