H.Headline
Chaos

THE EDITION ARCHIVE

A front page, saved in time.

Making AI Headlines · AM edition

Published · New York time

Permanent edition link →

10 pages · 23 stories

PAGE 1 OF 10

Amazon's Cloud Brings Sunshine, Its Forecast Brings the Umbrella

Strong cloud sales compete with a cautious holiday-quarter outlook and results below analysts' expectations.

Making AI Headlines newspaper, page 1: Amazon's Cloud Brings Sunshine, Its Forecast Brings the UmbrellaPage 1 of 10
Page 1 of 10. Read the same stories below.

Amazon's Cloud Shines While the Holiday Forecast Checks the Weather

Amazon reported strong cloud computing sales but gave a cautious forecast for the fourth quarter, leaving investors with a mixed picture ahead of the holiday shopping season. Its third-quarter results came in below Wall Street analysts' expectations. The contrast puts two important parts of the company in different light. Cloud computing supplies infrastructure customers use for services including artificial intelligence, while the fourth quarter covers the retail calendar's busiest stretch. Strong demand for computing can impress investors without settling their questions about sales across the wider company. The earnings report offers a corporate balancing act: celebrate the business that looks strongest, then choose the words around the next quarter with the care usually reserved for a holiday gift that might be returned. Amazon has a cloud to boast about, but its forecast leaves less room for a victory parade. For investors assessing AI infrastructure, sales are one part of the equation; the cost of building and running computing capacity is another. A strong cloud business does not automatically answer every question about the company's overall growth. Amazon's report gives Wall Street both a reason to admire the cloud and a reason to check the forecast. Apparently even a company that delivers almost everything cannot promise a perfectly sunny holiday season.

Share articleOpen article permalink →

Nvidia's Inner Circle Gets a Geopolitical Guest List

US intelligence agencies believe an executive in Nvidia chief executive Jensen Huang's circle is an informant for China, according to reporting by The New York Times cited by The Decoder. The belief attributed to intelligence agencies is not a court finding. The report places the people around one of the AI industry's most prominent leaders in a national security context. Nvidia's chips are central to computing systems used to develop and run AI, making the company a focus of international competition over advanced technology. That gives ordinary business relationships an unusually weighty setting. A technology executive's network can include customers, partners and acquaintances across many countries, while governments are increasingly interested in who has access to strategically important firms and their leaders. The claim concerns what US intelligence agencies believe about one executive. It does not establish that everyone in Huang's professional circle is under suspicion or that a particular business transaction has been compromised. Silicon Valley has long treated networking as a way to create opportunities. In this case, the same word comes with a distinctly less cheerful national security briefing. The industry's favorite advice, get to know the right people, has acquired an uncomfortable second question: who else is interested in them?

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 2 OF 10

Google's New Model Tops the Charts, But Its Audience Must Wait

Gemini 4 Carbon boasts coding and cybersecurity benchmarks, while US regulators examine an Nvidia deal.

Making AI Headlines newspaper, page 2: Google's New Model Tops the Charts, But Its Audience Must WaitPage 1 of 10
Page 2 of 10. Read the same stories below.

Gemini 4 Carbon Collects Benchmark Trophies Before Meeting Its Users

Google has announced Gemini 4 Carbon, a new AI model with leading benchmark scores in areas including coding and cybersecurity. Ars Technica describes it as essentially a slightly better version of Argon and says users cannot use it yet. The announcement captures an AI industry spectacle: a model can accumulate impressive public scores before the public gets a chance to put it to work. Benchmarks measure performance on specified tasks, so a strong result can help researchers and customers compare systems. They do not by themselves settle how a model will perform on every messy task a user might have. Coding and cybersecurity are consequential categories for a company selling AI tools to developers and organizations. A model that performs well on tests in those areas has a useful claim to make, even as users will ultimately care about the work it can handle for them. Google calls the model a significant leap. The competing description, a slightly better Argon, makes the same development sound less like a moon landing and more like a software update with excellent posture. Both put attention on progress, but they set very different expectations for the size of the step. For now, the scoreboard has arrived before the players can leave the tunnel. The rest of the evaluation will have to wait until people can actually use the model.

Share articleOpen article permalink →

Nvidia's Open-Source Ambitions Get a Closed-Door Review

The US Justice Department is looking into Nvidia's planned acquisition of Hugging Face, an open-source AI platform, as well as Nvidia's sales practices, according to The Decoder. The scrutiny puts a proposed deal involving a major chip company and a prominent AI platform before antitrust officials. Hugging Face is associated with sharing AI models and tools, while Nvidia supplies computing hardware used across the industry. A combination of those roles raises questions about how a major provider of infrastructure would fit with a platform built around access to shared resources. The review is not a finding that the deal or the sales practices violate antitrust law. It is an examination of a proposal and business conduct, with the eventual outcome still to be determined. For developers, the attraction of open tools is that they can be shared and used across a broad community. For regulators, the question is how the market works when an influential company proposes to bring an important platform into its orbit. Those are different perspectives on the same business relationship. The proposed marriage of AI hardware and an open-source platform now has another party at the table. The Justice Department is not there to catch the bouquet. It wants to read the agreement and ask about the guest list.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 3 OF 10

Wikipedia Asks Courts to Locate the Nuance Button

Wikimedia draws a line between AI training and reproduction, while a Kentucky administrator obtains an exemption from workplace AI.

Making AI Headlines newspaper, page 3: Wikipedia Asks Courts to Locate the Nuance ButtonPage 1 of 10
Page 3 of 10. Read the same stories below.

Wikimedia Wants AI to Read the Web Without Reciting It

The Wikimedia Foundation is asking courts to protect AI training on copyrighted works while not protecting chatbots that reproduce those works for users. The operator of Wikipedia says an overly rigid ruling in major copyright cases could threaten the open web. Its position separates two uses that can get bundled together in arguments over AI. Training a model on material is one question; a chatbot repeating that material to a user is another. Wikimedia wants the courts to make room for the first without giving the second an automatic pass. The foundation also warns that a sweeping ruling could deprive Wikipedia contributors of useful AI tools or endanger the online environment on which their work depends. Wikipedia itself is a prominent example of a project built around making information accessible, while relying on people to assemble and maintain it. The cases involve competing interests in creative work, AI development and access to information. Wikimedia's proposal does not settle where a court should draw every boundary, but it gives judges a more precise distinction than simply declaring all AI uses good or bad. An encyclopedia famous for footnotes is asking the legal system for a ruling with room for one. In a debate that often offers two enormous buttons marked yes and no, Wikimedia has submitted a request for a third: it depends.

Share articleOpen article permalink →

Kentucky Administrator Finds a Conscience Clause in the AI Rollout

David Durman, a University of Kentucky administrator, obtained a religious exemption from using AI at work. The case illustrates that a religious accommodation does not necessarily require belief in a god. The decision brings a workplace question into an area usually discussed in product announcements and productivity claims. As organizations introduce AI tools, some employees may object to using them for reasons that go beyond whether they like the software. Durman's accommodation shows one way such an objection has been addressed in a particular workplace. An exemption is specific to the person and circumstances involved. It does not mean every objection to AI will qualify, nor does it establish a general right for all workers to opt out. It does show that a request involving a religious belief need not fit a narrow assumption about belief in a deity. For employers, the case sits at the intersection of technology policy and existing accommodation practices. A tool can be marketed as a universal aid, but a workplace still contains individuals whose beliefs and responsibilities differ. The AI rollout promises to make every office task smoother. At least one office has discovered that its most important document may be the one explaining why a particular employee will not be using the tool at all. Conscience, unlike a software update, does not come with a universal install button.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 4 OF 10

Amazon Dreams of a Trillion, Microsoft Offers an Exit

Andy Jassy ties AWS ambitions to continued spending, while Microsoft offers voluntary buyouts to roughly 60,000 employees.

Making AI Headlines newspaper, page 4: Amazon Dreams of a Trillion, Microsoft Offers an ExitPage 1 of 10
Page 4 of 10. Read the same stories below.

AWS Sets a Trillion-Dollar Table and Keeps Ordering Infrastructure

Amazon chief executive Andy Jassy says AWS could become a $1 trillion revenue business, but reaching that ambition will require continued heavy capital spending. Amazon wants a leading position in AI cloud computing, and Jassy's pitch puts the cost of building capacity alongside the enormous revenue goal. Capital expenditure includes the long-term investments needed to expand computing infrastructure. For a cloud provider, that means spending before the full return is known. The scale of the ambition makes the scale of the bill an essential part of the story, not a small footnote underneath it. Businesses that rent computing power are potential customers for that capacity, including organizations developing and using AI. Whether demand grows enough to justify the investment is a business question that cannot be answered by the size of the target alone. Jassy's trillion-dollar figure is an aspiration, not a report that AWS has already reached that level. It is also a reminder that the competition to supply AI computing is expensive long before anyone gets to celebrate a milestone. The cloud industry's preferred growth chart points sharply upward. Its spending chart has apparently received the same instructions. Amazon's pitch is that the second line helps produce the first, an attractive story as long as customers eventually fill the very large table being set for them.

Share articleOpen article permalink →

Microsoft Makes Its Smaller Workforce Offer Voluntary

Microsoft is offering voluntary buyouts to around 60,000 employees across the United States, United Kingdom, Canada and other locations, according to The Verge. The offer comes as chief executive Satya Nadella has discussed AI's potential effects on the company's workforce. A voluntary buyout gives an eligible employee a choice about whether to accept an offer to leave. It is different from saying that all those offered the package will depart, or that AI has already replaced a particular number of employees. The size of the offer nevertheless makes it a significant workforce development at one of the world's largest technology companies. The Verge assesses the move as a sign of how AI may change the size of Microsoft's workforce. That is an interpretation of the broader direction, not a count of jobs eliminated by a specific AI system. The business tension is clear: Microsoft sells tools meant to make work more efficient while also managing its own workforce. Employees must make their own decisions about an offer, and the company must still organize the work that remains. Corporate language has a gift for making a large change sound like a quiet invitation. In this case, the invitation really is voluntary, but it arrives at a moment when the company's vision of an AI-enabled workplace makes the future organization chart a topic of considerable interest.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 5 OF 10

Reddit Locks the Door While Rivian Lends Volkswagen Its Voice

A login-led Reddit responds to changing search traffic, and a car partnership expands into AI assistants.

Making AI Headlines newspaper, page 5: Reddit Locks the Door While Rivian Lends Volkswagen Its VoicePage 1 of 10
Page 5 of 10. Read the same stories below.

Reddit Tells the Open Web to Please Sign In

Reddit says it is moving toward a primarily logged-in experience as traffic from search changes and AI reshapes how people find information online. The company announced the shift on its third-quarter earnings call. For years, people looking for human answers have added Reddit to their search queries, sending visitors from search engines to conversations on the site. That route has become less dependable as AI-powered search changes the relationship between search results and the sites supplying information. The move is intended in part to protect Reddit's data from AI scraping and support licensing deals with model providers. It could also help the company increase revenue as it faces declines in daily active users in its largest markets, according to TechCrunch. A more login-centered site changes the bargain for someone who arrives looking for a quick answer. It can give Reddit more control over access to conversations that its users create, while adding a step between a search result and the discussion. The business logic is understandable: a platform cannot pay its bills with the warm feeling that its answers are useful somewhere else. Still, there is a tidy reversal in a site famous for improving search results asking searchers to identify themselves before coming inside. The internet spent years typing Reddit into the search box. Reddit would now like the internet to type in a password, too.

Share articleOpen article permalink →

Rivian Lends VW an AI Voice and Keeps the Fee Quiet

Rivian is licensing elements of its AI voice assistant technology to Volkswagen Group as the companies expand their $5.8 billion joint venture to include AI assistants. Rivian disclosed the expansion in its third-quarter investor presentation. The company did not disclose licensing fees, payment terms or which Volkswagen vehicles will use the technology. It also did not say whether the value of the joint venture would expand as a result. Those details leave the financial effect of the new arrangement open. The assistant recently rolled out to Rivian vehicles. It could make its Volkswagen debut in the low-cost ID EVERY1, a small four-door hatchback slated to be the first Volkswagen vehicle to use Rivian software and electrical technology. That possibility is not a confirmed assistant launch for the vehicle. The arrangement extends a partnership already focused on vehicle technology into a feature drivers may interact with directly. For automakers, sharing software can offer a way to bring features across brands without each company independently building every component. The assistant's job is to respond when someone asks a question. Its business arrangement is more reserved. Rivian has explained that the technology is traveling to Volkswagen, but the price of the trip remains firmly in the glove compartment.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 6 OF 10

OpenAI Raises the Limit and AI Investors Visit the Front Door

Paid-plan quotas double after a reversal, while OpenAI-backed Thrive Holdings agrees to buy ADT.

Making AI Headlines newspaper, page 6: OpenAI Raises the Limit and AI Investors Visit the Front DoorPage 1 of 10
Page 6 of 10. Read the same stories below.

OpenAI Cuts the Cut and Doubles the Allowance

OpenAI is doubling usage limits across its paid plans from Pro 100 to Pro 500, after reversing a planned cut to the Pro 200 quota following user pushback. The company says more efficient models, including Terra and Sol, should increase usable capacity further. The change comes ahead of Anthropic's planned initial public offering as the companies compete for business customers. Usage limits matter to subscribers who depend on AI tools for regular work: a higher allowance gives them more room before they reach the plan's ceiling. The sequence is especially revealing. A proposed reduction met resistance, the reduction was reversed, and paid-plan limits are now being doubled. It is a vivid example of how a technical-sounding quota is also a product decision that customers can compare across providers. The promised additional capacity from more efficient models is a company expectation. How much benefit individual users see will depend on the tools and plans they use. The announced doubling, meanwhile, changes the terms on which paid customers can use the service. AI companies often present model efficiency as an abstract engineering achievement. Subscribers experience it as the difference between getting more work done and being told to come back later. This time, the limit went on a remarkable journey: down in the plan, back up after complaints, and then considerably higher. The customer feedback form has rarely looked so powerful.

Share articleOpen article permalink →

Thrive Pays $15.3 Billion to Give Home Security an AI Pitch

OpenAI-backed Thrive Holdings is buying home security company ADT in a deal valued at $15.3 billion, with plans to bring AI to the business. The acquisition joins an established alarm and monitoring brand with an investor looking to expand the use of artificial intelligence. ADT's business gives the deal a practical setting: security services are tied to homes and other physical properties, rather than simply to a chatbot on a screen. The stated plan is to bring AI into that business. The value of the proposed technology will be judged as products and services take shape. A plan to use AI is not itself a description of a particular alarm, monitoring feature or change to a customer's service. The purchase price, meanwhile, puts a substantial figure on the transaction before those details are clear. The deal also shows how far AI investment ambitions extend beyond the companies building models. An existing security business has customers, equipment and a recognizable brand, while its prospective owner has a plan to add a technology that has become a central selling point across the market. Home security has traditionally made a simple promise: pay attention when something needs attention. AI adds a much grander vocabulary to that promise. The ultimate product still has a rather down-to-earth test: whether the system is useful to the people relying on it.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 7 OF 10

A Coin Advises the Experts and Meta Organizes the Closet

An Anthropic study examines AI influence on judgment, while Meta launches a fashion app linked to Instagram.

Making AI Headlines newspaper, page 7: A Coin Advises the Experts and Meta Organizes the ClosetPage 1 of 10
Page 7 of 10. Read the same stories below.

Experts Follow a Coin's Advice and Call It Their Own

In an Anthropic study, 50 experts used an AI agent whose preferences were set by a coin toss. On judgment calls, the experts followed the agent's view up to 86 percent of the time, and most believed the conclusions were their own, according to the study. The result focuses on a subtle form of influence. The agent's preference was assigned randomly, but it was presented through an AI system. The study reports that experts often adopted its view even when the preference itself had been determined by a coin. The figure is an upper reported rate for judgment calls in the study, not a claim that every expert followed the agent 86 percent of the time or that all professional decisions work this way. The group included 50 experts, so its results are evidence from that study rather than a census of everyone who uses AI. Anthropic is itself an AI company, giving it a direct interest in how people understand and use agents. The finding raises a question about a tool that can present a position fluently: an answer can influence a judgment even when the preference behind it was assigned at random. The experts brought professional knowledge to the exercise. The coin brought two sides. In the study's most striking result, the coin still managed to become a remarkably successful colleague.

Share articleOpen article permalink →

Meta Turns Instagram's Past Outfits Into a Closet With Opinions

Meta has launched Style Sketch, an AI fashion app that can connect to Instagram and build a virtual closet from photos users have already shared. People can also add items manually by taking a photo, recording a short video or uploading images from their camera roll. The app takes material that may already exist in a user's social media history and gives it another job. Instead of serving only as a record of what someone wore, a photo can help populate a digital wardrobe. Manual additions offer another route for clothes that have not appeared in an Instagram post. That makes the product an example of a platform strategy: turn an existing collection of content into a new service. The convenience for users is having wardrobe information gathered in one place; the business advantage for Meta is another use for material and connections already associated with its ecosystem. The launch also makes account connections and the images a user chooses to provide part of the experience. A virtual closet is only as representative as the items that make it inside, whether they arrive from Instagram or a newly captured image. Fashion has always involved looking back at what worked. Meta's version can begin with an archive of photos that the user already posted. The closet may be virtual, but its first task is very real: sorting through last summer's outfits without making anyone open a drawer.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 8 OF 10

Factory Robots Face a Border Check as Gemini Takes Over the Filing

Trump orders trade talks on imported industrial robots, and Google adds file editing inside Gemini.

Making AI Headlines newspaper, page 8: Factory Robots Face a Border Check as Gemini Takes Over the FilingPage 1 of 10
Page 8 of 10. Read the same stories below.

Trump Gives Imported Factory Robots a National Security Review

President Trump has declared imports of foreign-made factory robots a national security threat, ordered trade talks and threatened tariffs. The action puts industrial automation into a trade dispute with potential consequences for manufacturers that buy equipment across borders. Factory robots are tools used in production, but they are also part of the infrastructure on which manufacturers rely. The administration's move treats where those machines come from as a security and trade question, not just a purchasing decision. A threatened tariff is not the same as a tariff already imposed. The ordered talks and the possibility of duties nevertheless create uncertainty for businesses considering imported equipment. The eventual effect would depend on the details of any measures and the outcome of negotiations. Manufacturers weighing automation already face questions about the cost of machines and how to integrate them into production. Trade policy can add another variable to those decisions, particularly when equipment and components cross borders before reaching a factory floor. The policy offers a modern version of an old trade argument. A machine can perform the same task regardless of its birthplace, but its origin may determine how welcome it is in an American factory. The robot does not get a vote in the negotiations, although its buyer will have a keen interest in the final price.

Share articleOpen article permalink →

Gemini Invites Your Documents Into the Chat and Gives Them Homework

Google has rolled out file editing in Gemini, allowing users to generate, preview and edit formats including PDFs, Google Docs, Google Sheets, slide decks, Markdown, text files and code within the chat. The new capabilities are available to all Gemini users. There is an important platform distinction: PDF previews and editing work on both the web and mobile, while the other file formats have more limited availability. The feature brings document work closer to the conversation where a user asks for it, rather than requiring every result to be copied into a separate workspace. For someone drafting a report or working through a spreadsheet, the appeal is a shorter route from an instruction to an editable file. It also makes the chat window responsible for more of the workflow, from creating material to previewing changes. An editable result still needs a person's attention, especially when the document will be shared or used for a consequential decision. A neat preview can make a change easy to see, but it cannot guarantee that every change is the one the user intended. The office has spent years moving paperwork from desks to screens. Now Google wants to move more of it into a conversation. The chatbot was introduced as something that could help with a document; it has progressed to inviting the document into its own office and offering it a chair.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 9 OF 10

The Pentagon's Surprise AI Colleague Skips the Onboarding

Sources describe an agent on a defense network, while voice apps converge and Mistral opens model weights.

Making AI Headlines newspaper, page 9: The Pentagon's Surprise AI Colleague Skips the OnboardingPage 1 of 10
Page 9 of 10. Read the same stories below.

A Personal Agent Finds Two Months of Work on a Defense Network

An Accenture employee's personal Claude agent ran on a Department of Defense network for two months before it was identified and the vulnerability could be patched, sources told WIRED. The account puts an employee's personal AI tool inside a sensitive government system. The episode raises a question for organizations using sensitive networks: how do they identify AI tools that have access to their systems? The reported two-month interval puts the focus on both access and the time needed to find it. AI agents are promoted for their ability to carry out tasks on a user's behalf. On a defense network, however, the important question is not only what an assistant can do, but whether its access has been identified and managed. The Pentagon's surprise colleague managed an extended stay before the relevant vulnerability could be patched. For an industry selling agents as tireless helpers, that is an unusually awkward performance review.

Share articleOpen article permalink →

Granola and Wispr Compete to Become the Same Meeting Companion

Granola is adding dictation to its AI meeting notetaker, after Wispr added meeting notes to its dictation app. The two products are approaching the same set of tasks from opposite directions. Dictation turns spoken words into text, while a meeting notetaker helps capture what happened in a conversation. Combining the features gives each company a way to compete for more of the user's voice-based workflow. The rivalry is a neat reversal: the notetaker wants to take dictation, and the dictation app wants to take notes. Both are betting that people would rather use one assistant for related work than switch between tools. For the person attending the meeting, the contest may be less dramatic than the product roadmaps suggest. The immediate goal is still to leave with usable notes and fewer words to type afterward. The assistants, meanwhile, are busy agreeing that each other's features look excellent.

Share articleOpen article permalink →

Mistral Opens a Trillion-Parameter Model and Sells Control

Mistral has released weights for a trillion-parameter AI model aimed at cybersecurity. The company says the release can give organizations a kind of control over cyber defenses that closed models cannot offer. Open weights give organizations another way to work with a model, rather than relying entirely on access through a closed provider's interface. For a security team, control over how a model is used can be an important consideration, although the company's comparison remains its own claim. The release adds an option to a field in which model capabilities and deployment decisions both matter. A capable tool still needs to be evaluated for the work an organization wants it to perform. Mistral's pitch turns openness into a security feature: hand customers more control over the model and let them decide how to put it to work. In an industry fond of keeping the machinery behind a curtain, the company is advertising a closer look at the machinery itself.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑

PAGE 10 OF 10

AI Gets Five Values, More Ads and a Shopping List

OpenAI updates its model guidance, Alexa handles lists, Google weighs search revenue, and Dream funds a construction model.

Making AI Headlines newspaper, page 10: AI Gets Five Values, More Ads and a Shopping ListPage 1 of 10
Page 10 of 10. Read the same stories below.

OpenAI Gives Its Chatbot Five Values and a Long To-Do List

OpenAI has updated its Model Spec with five core values intended to guide more capable AI toward sound judgment, human agency and faithfulness to human intent. The document sets out the company's desired direction for model behavior. A published set of principles gives users a way to examine what OpenAI says it wants its systems to do. The harder task is making those principles work across the many situations a model encounters. Corporate values are often displayed on a wall. OpenAI's must contend with a chatbot expected to answer everything from a complex question to a request for lunch ideas. The principles have an unusually busy workday ahead.

Share articleOpen article permalink →

Alexa Adds List Management to Its Retail Résumé

Amazon has updated Alexa to help users create and manage lists, add or edit items, manage subscriptions and update delivery instructions. The voice assistant is taking on more of the routine work around shopping and orders. A single spoken request can be more convenient than opening several screens to track household tasks. It also places more of those tasks within Amazon's ecosystem, where a list can sit comfortably next to the business of buying what's on it. Alexa began as a voice assistant for everyday requests. Its latest job description sounds rather more like a household administrator, with a particular interest in making sure the packages know where to go.

Share articleOpen article permalink →

Google's AI Answers Need More Ads to Pay Their Way

Google says AI search makes less money and is responding with more advertising, as investors assess its third-quarter earnings. Ars Technica reports that negative cash flow reached $9.06 billion and the stock fell after the results. The business challenge is to make a newer search experience earn enough while users change how they seek information. Ads are Google's familiar revenue tool, even when the presentation of answers is changing. The company has given search a conversational makeover. Its financial accessory is more traditional: another opportunity to put an advertisement near the answer. The format may be new, but the sales department still recognizes the furniture.

Share articleOpen article permalink →

Dream Raises $17 Million to Make AI Learn Building Codes

Former Palantir executives have launched Dream with $17 million from a16z to build AI world models for construction. The company's model is grounded in physics and construction standards and trained on more than 100 million square feet of building data from US permit archives. The goal is a system that works with the constraints of the built environment, rather than merely producing plausible language about a project. Permit data gives the model a connection to records of actual buildings. Construction is a demanding audience for an AI pitch. A confident answer is pleasant, but walls and permits have a stubborn habit of requiring the details to work.

Share articleOpen article permalink →

Sources from the news desk

Satire. Not factual reporting. Read the original reporting from the linked publishers.

Back to page 1 ↑