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

AI Offers Workers a Stake in the Plot

Anthropic plans a fund backed by 5 percent of its shares, including possible payments to workers facing AI disruption.

Making AI Headlines newspaper, page 1: AI Offers Workers a Stake in the PlotPage 1 of 10
Page 1 of 10. Read the same stories below.

Your Replacement May Come With a Benefits Brochure

Anthropic plans to commit 5 percent of its shares to a public benefit fund that will explore direct payments to workers at risk of losing jobs to AI. The fund is structured as a company-owned LLC, leaving Anthropic in control as a possible IPO approaches. That makes its governance as important as its generosity. A share commitment is not the same thing as a check arriving when rent is due. The practical questions concern who qualifies, who decides, and how a corporate asset becomes useful help. Workers need support on household calendars, which rarely include a convenient box marked eventual liquidity. Still, the proposal acknowledges something the usual productivity presentation tends to leave outside the frame: efficiency gains can have very specific mailing addresses, while their costs land elsewhere. The comic business model is almost elegant. First promise software that changes employment forever, then offer an ownership-linked umbrella for the weather. Whether that umbrella opens remains a governance question, not a branding question. A public benefit fund will ultimately be assessed by what reaches the public, rather than how reassuringly the word benefit sits beside the company logo.

Share articleOpen article permalink →

Meta Deletes the Part That Made Publishers Reach for the Locks

Meta is removing language from its commercial terms that raised concerns it could scrape websites, including news sites, to train AI. The change follows backlash over wording that appeared to give the company broad permission to use content. For publishers, the stakes are control over material that costs money to produce. Terms pages are an unusual venue for a public negotiation. One side has reporters, editors, servers, and payroll. The other side has a sentence that can arrive looking like routine maintenance and leave everyone checking whether the furniture is still theirs. Removing the language addresses the immediate wording dispute. It does not make the larger tension disappear: AI developers want abundant material, while publishers want meaningful authority over its use. Neither interest becomes simpler because the agreement is displayed in a very small font. The useful lesson is that legal text can function like product design. A broad clause changes the relationship even when almost nobody reads it before clicking. Publishers have now demonstrated the one engagement metric a terms page rarely seeks: an audience alert enough to insist on an edit before the next update.

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

The Future Meets City Hall and a Very Large Invoice

New York's mayor calls for public control over AI as Anthropic and Nscale reportedly discuss another $45 billion compute deal.

Making AI Headlines newspaper, page 2: The Future Meets City Hall and a Very Large InvoicePage 1 of 10
Page 2 of 10. Read the same stories below.

Mamdani Asks Residents Not to Click Accept on Destiny

New York Mayor Zohran Mamdani urged residents Monday night to reject the idea that an AI takeover is inevitable and demand rules that protect the public. He argued that federal authorities have abdicated responsibility. His intervention places AI within ordinary political debate rather than treating it as weather arriving from a server farm. That distinction matters. A technology can improve quickly without every business decision made in its name becoming unavoidable. Deployment, public protections, and accountability involve choices, even when the sales presentation uses a timeline with only one arrow. For residents, the useful question is less whether the future is coming than who gets to set its terms. Regulation can involve tradeoffs, including how to preserve useful tools without asking the public to absorb every experiment's downside. The inevitability pitch has an appealing convenience for its sellers: nobody has to defend a decision if it can be described as gravity. Mamdani is asking people to inspect the pulley system. City politics may not settle national AI policy, but it can challenge the notion that democracy should wait politely in the lobby while the product team finishes deciding what happens next.

Share articleOpen article permalink →

Another $45 Billion, Subject to the Small Matter of Agreement

Anthropic and infrastructure provider Nscale are reportedly discussing another $45 billion compute deal, a little over a month after announcing one of the same size. Bloomberg reported the talks, citing people familiar with them, who cautioned that negotiations remain underway and details could change. The prospective agreement is not a completed purchase. The scale highlights how AI competition increasingly depends on securing computing capacity, not just producing a persuasive demonstration. A model may live in a chat window, but the business behind it requires infrastructure commitments with considerably less charming loading animations. For customers and business partners, the practical issue is whether capacity, demand, and financing develop together. A large agreement signals ambition; it does not by itself establish how efficiently the resulting resources will be used. At this size, the familiar office question of whether someone really needs another monitor has evolved into whether someone really needs another $45 billion of compute. The underlying assessment is still recognizable: what work requires the equipment, on what schedule, and at what cost? The number may be enormous, but utilization remains stubbornly resistant to being impressed by a press release.

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

Trust Gets a Visitor Badge

Hugging Face's policy chief questions embedded evaluators, while Microsoft urges businesses to watch AI agents closely.

Making AI Headlines newspaper, page 3: Trust Gets a Visitor BadgePage 1 of 10
Page 3 of 10. Read the same stories below.

An Outside Evaluator Is Not a Safety Spell

Hugging Face's policy chief says embedding outside evaluators inside AI organizations will not be an easy safety solution. The approach puts ostensibly neutral observers within labs and has gained attention as concerns about AI harms grow. Its appeal is straightforward: let someone other than the builder inspect what is happening. The harder question is what makes that inspection independent and useful. Physical proximity is not the same as authority. An evaluator can sit close to the machinery while still depending on someone else for access, time, and the ability to communicate findings. Good oversight needs more than a chair near the engineering team. Its value depends on whether meaningful scrutiny can survive commercial urgency, institutional habits, and disagreements about what deserves attention. Those are design questions for governance, not reasons to abandon scrutiny. The tempting corporate shortcut is to treat an observer as a decorative safety feature, like a tiny lighthouse placed on the conference table. It looks excellent in the diagram. Whether ships actually avoid the rocks requires a separate assessment. The policy chief's warning is a reminder that adding a person to an organization chart does not automatically add accountability to the organization.

Share articleOpen article permalink →

The Digital Intern Needs Fewer Keys

Microsoft is urging enterprises to closely monitor AI agents used by employees, treating them as potential insider threats rather than automatically trustworthy assistants. Its warning focuses attention on how businesses supervise agents and on controls that do not simply depend on the model providers. The office helper now needs an access review. The issue is practical. Software that can act on a user's behalf can also carry the consequences of that user's permissions. Convenience grows when an agent can reach more systems, but so does the importance of limiting what it can do and observing what it actually does. That changes the procurement conversation. A polished demonstration of an agent completing a task is only the opening act. Businesses also need to assess permissions, oversight, and what happens when confident activity does not match the intended instruction. The old office rule was not to give the new hire every password before lunch. The updated version apparently requires explaining that fluent sentences are not a background check. An agent can be useful without being trusted with the entire building. Indeed, a helper that accepts boundaries may be more valuable than one whose main qualification is sounding delighted to have administrator access.

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

The Cloud Discovers Gravity

Rising US electricity demand accompanies a fossil fuel buildout, while CoreWeave's debt draws scrutiny.

Making AI Headlines newspaper, page 4: The Cloud Discovers GravityPage 1 of 10
Page 4 of 10. Read the same stories below.

The Future Plugs Into Some Familiar Fuel

The United States is building more fossil fuel infrastructure than it has in a decade, according to a new report, while renewable energy growth slows. Electricity demand is also rising at its fastest pace in more than a decade, with data centers, factories, and electric vehicles adding to the load. AI's expansion belongs in that larger power discussion. The stakes are physical rather than rhetorical. Computing needs electricity, and electricity systems must match demand with available supply. Calling a service cloud-based does not lift its energy requirements into the sky, however airy the website's gradient may be. The report also complicates simple explanations. Data centers are part of the demand increase, not its only cause. An honest assessment has room for industrial growth, transportation changes, and choices about how new power gets built. The technology sector likes to describe progress as a leap beyond the old economy. The grid keeps replying with a request for a connection form. As new digital services expand, their environmental consequences depend partly on decisions made far from the chat interface. The future can arrive in a sleek browser window and still need a very traditional conversation about fuel, infrastructure, and who pays.

Share articleOpen article permalink →

CoreWeave's Bonds Add a Less Cheerful Progress Bar

CoreWeave's bonds slipped into distressed territory by a common market definition: investors demanded a yield premium of 1,000 basis points over government bonds. That is a signal of heightened concern in the debt market, not a declaration that the company has failed. It matters because financing sits underneath the AI industry's infrastructure ambitions. A computing business can have strong demand narratives and still face difficult questions about the cost of borrowing. Debt investors assess repayment risk, not merely whether a technology is impressive. Their spreadsheets are notoriously unmoved by a demo that writes a delightful poem about server racks. For the broader sector, the development draws attention to how infrastructure gets funded. Capacity commitments, equipment spending, and the terms of financing have to coexist in the same business, even when they occupy different slides in a presentation. The useful comparison is between operating ambitions and financial obligations, not between enthusiasm and pessimism. Neither a gloomy bond price nor an exuberant keynote settles the whole assessment. But debt has a distinctive talent for interrupting grand narratives: it asks for its money on a schedule. The future may be transformative; the interest payment remains an appointment.

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

AI's Workforce Moves, Its Credentials Wander

Disney's bid includes jobs for most Character.AI employees, while OpenCode users face stolen API keys and unwanted charges.

Making AI Headlines newspaper, page 5: AI's Workforce Moves, Its Credentials WanderPage 1 of 10
Page 5 of 10. Read the same stories below.

Disney's AI Deal Comes With an Employee Entrance

Disney has agreed to hire the majority of Character.AI's workforce, and a filing says the vast majority of employees offered jobs accepted. The arrangement is a stalking horse bid, meaning another company could still submit a higher offer. The employment plans are therefore part of a process that remains open to competition. For the AI industry, the development illustrates how a company's people can become central to a transaction. Software may be the visible product, but the expertise needed to build and operate it does not automatically transfer with a logo or a collection of computers. For employees, an accepted offer is a more concrete development than another sweeping prediction about the future of intelligent entertainment. Yet the bidding structure matters: a proposed destination and a completed process are not the same thing. There is a particular modern-business elegance to technology sold as endlessly scalable arriving at the old question of where everyone will work on Monday. The cartoon characters may be digital, but career decisions remain stubbornly human. This is also why the paperwork deserves attention. In a sector fond of announcing new eras, a conditional transaction still has to proceed through the extremely analog business of offers, acceptances, and competing bids.

Share articleOpen article permalink →

OpenCode's Unwanted Feature: Someone Else's AI Bill

Hackers breached part of OpenCode's online platform, stole some users' API keys, and ran up charges on their AI accounts. OpenCode is an open source coding agent that has become an alternative to Claude Code and Cursor. The incident shows how access credentials can turn a development tool into a route to somebody else's paid services. The reporting also says AI tools consulted for help repeatedly refused to examine the malicious code. That creates an awkward distinction between refusing to assist wrongdoing and refusing to inspect evidence of wrongdoing. A safety boundary has to account for the purpose of an investigation, not merely the alarming appearance of the material. For users, the central issue is not which assistant sounds smartest. It is what an exposed key permits, how abuse becomes visible, and how quickly access can be contained. Those questions belong beside feature comparisons, not underneath them. The industry's dream is an assistant that handles the tedious parts of software work. The less marketable reality is that tedious security work remains essential. A tool that can explain a complicated program still needs a credential system that does not let strangers put their computing errands on the customer's tab.

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

Your Personal Assistant Would Like Some Runway

Tolan raises another $30 million for its companion app, while Replika's founder launches personal assistant Wabi.

Making AI Headlines newspaper, page 6: Your Personal Assistant Would Like Some RunwayPage 1 of 10
Page 6 of 10. Read the same stories below.

Tolan Finds Investors for the Business of Being There

AI companion app Tolan has raised another $30 million after attracting more than 250,000 active users. Its cartoon-style companion is designed to help people manage daily life, relationships, emotions, and goals. Investors in the round include Union Square Ventures, Khosla Ventures, and Homebrew, and the financing doubles the company's total capital raised. The business proposition combines a familiar desire for support with an unfamiliar delivery system. A character on a screen can be readily available, but availability alone does not establish the quality of its guidance. Growth and fundraising measure different things from whether an interaction is genuinely helpful. For users, the useful assessment concerns boundaries, expectations, and whether the app supports their own judgment. For the company, the challenge is building a durable service without turning every moment of uncertainty into another reason to keep chatting. Venture capital has discovered that the daily business of being a person contains a formidable number of recurring tasks. The comic possibility is a life coach whose own life goal is extending corporate runway. The substantive question is simpler: can a companion help users accomplish something beyond maintaining the companionship? A friendly character is an interface; usefulness still has to happen outside the interface.

Share articleOpen article permalink →

Wabi Arrives After a Twelve-Year Loading Screen

Replika founder Eugenia Kuyda has launched Wabi, a new startup offering a personal AI assistant that is free and available to download. The product builds on an idea she pitched at Disrupt 12 years ago. Its arrival illustrates how a long-standing software ambition can gain another opportunity as AI capabilities change. The personal assistant remains one of technology's most durable promises. It suggests fewer chores, less searching, and a little relief from managing the systems that were previously sold as labor-saving. Each generation inherits both the appeal and the burden of proving that an assistant reduces work rather than creating a new interface to supervise. For users, free availability lowers the initial barrier to trying the product. It does not answer the more interesting question of whether the tool fits real routines. The test is not how futuristic assistance sounds, but whether it makes an ordinary task easier. Twelve years is a long interval in launch language, where next quarter is often treated as the distant frontier. It is a shorter interval in the history of wanting somebody else to organize things. Wabi enters a market where the dream is familiar and the useful details will do the persuading.

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

Competition Arrives With Chips and Smaller Bills

Tenstorrent raises $1.5 billion to stay independent, while MiniMax reports stronger benchmark results at lower cost.

Making AI Headlines newspaper, page 7: Competition Arrives With Chips and Smaller BillsPage 1 of 10
Page 7 of 10. Read the same stories below.

Tenstorrent Buys More Time to Be Tenstorrent

AI chip company Tenstorrent has raised $1.5 billion at a $6.2 billion valuation after attracting acquisition interest earlier this year. The Nvidia challenger is using the financing to remain independent and scale its open-weight AI chip ambitions. The round adds resources to a competition where building an alternative requires considerably more than declaring one. For customers, another well-funded chip developer offers the prospect of more choice. But capital raised and useful competition are different milestones. Hardware must fit the software, workloads, and operational needs of organizations that have good reasons to dislike expensive surprises. That makes independence both an opportunity and an obligation. Tenstorrent retains the chance to pursue its own direction, while also retaining the work required to turn that direction into a durable business. An acquisition offer and a financing round solve different problems. In ordinary life, remaining independent may mean declining help carrying a couch. In AI hardware, it can involve raising $1.5 billion. The scale is comic; the competitive question is serious. Buyers benefit when alternatives prove themselves through practical performance and usability, not simply when the industry adds another impressive valuation to the scoreboard. The money funds the attempt, rather than finishing it.

Share articleOpen article permalink →

MiniMax Makes the Benchmark Table Bring a Calculator

MiniMax's M3.1 Flash scored 42 on the Artificial Analysis Intelligence Index, a 13-point improvement over its predecessor, at half the cost. It also reached 73.7 percent on the DeepSWE v1.1 coding benchmark for under 30 cents per task, matching DeepSeek V4.1 Flash 1017 on reported performance and price. The results add pressure to global competition over AI cost and capability. For American developers comparing models, the practical attraction is not national scoreboard theater. It is the possibility of completing a particular workload with less spending. A benchmark is useful evidence, but its relevance depends on how closely the test resembles the work someone actually needs done. Lower prices can widen access and make experimentation cheaper. They also push providers to explain what customers receive for a premium beyond a more prestigious name in the model selector. The industry's old magic trick was making software produce an answer. The newer trick is making the invoice small enough that someone can afford to ask again. Still, a benchmark table is not a universal purchasing decision. Performance, reliability, and fit remain part of the assessment, even when the cost column is doing its best to steal the meeting.

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

More Frames, Fewer Safe Previews

Kling expands its video generator, while fear of AI game clones makes some independent developers wary of showing unfinished work.

Making AI Headlines newspaper, page 8: More Frames, Fewer Safe PreviewsPage 1 of 10
Page 8 of 10. Read the same stories below.

Kling Gives the Prompt a Longer Production Meeting

Kling AI has released version 4.0 of its video generator, supporting clips up to 30 seconds, as many as ten keyframe images, and prompts of up to 8,000 tokens. The release also adds stereo audio, multilingual lip sync, and a lighter Flash variant. Reported comparisons with Wan 3.0 and Seedance 2.5 place it among competitors rather than clearly ahead. The practical development is greater room to specify what a sequence should contain. Keyframes and longer prompts can give creators more ways to communicate intent, while audio and lip sync address parts of a clip that pictures alone cannot carry. More input capacity, however, is not automatically less work. Creative control depends on whether the system follows useful directions consistently, not simply on how many directions it accepts. An enormous prompt box can become a screenplay, a production memo, and a small cry for help. For video makers, the sensible comparison is task by task: what can be directed, revised, and used? Thirty seconds is enough time for an idea to land, but also enough time for inconsistencies to become noticeable. The feature list expands the toolkit. The user's taste and patience still have speaking roles.

Share articleOpen article permalink →

Indie Developers Confront the Cost of Showing Off

Fear of AI-generated game clones is making some independent developers reluctant to show their work publicly. A delayed Steam Next Fest demo illustrates the concern: a preview intended to attract players can also reveal ideas to people looking to imitate them. For small studios, visibility is both a business necessity and a source of exposure. The tension is especially awkward because a demo is supposed to reduce uncertainty. Players learn whether a game interests them, and developers learn whether their pitch connects. If showing the work feels risky, both sides lose part of that exchange. The issue is not that every resemblance proves copying, or that every AI tool has the same use. It is the incentive created when imitation becomes easier while original development still requires time, judgment, and sustained effort. Marketing then starts to resemble opening a bakery while wondering whether the window display includes the recipe. The comic nightmare is a launch calendar in which the knockoff appears before the thing it resembles. The substantive concern is quieter: creators may become less willing to share early work at all. An ecosystem can gain more generated products while losing some of the openness that helped distinctive games find their audience.

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

AI Comes Home, Sometimes to Remove the Features

Smart TV owners use agents against bloatware as Nvidia promotes locally run models and Illinois' governor weighs AI's stakes.

Making AI Headlines newspaper, page 9: AI Comes Home, Sometimes to Remove the FeaturesPage 1 of 10
Page 9 of 10. Read the same stories below.

The Smartest TV Is Apparently the One That Stops Talking

Some smart TV owners are using AI agents to strip away bloatware, frustrated by ads, tracking, and slow interfaces. The aim is not a more elaborate television experience, but greater control over the device they already own. It is an unusually clear consumer verdict: advanced software is being recruited to undo other software's advances. The desired breakthrough is a television that behaves like a television. The practical tension concerns who defines improvement. Manufacturers can add connected features, while owners may value speed, simplicity, and fewer interruptions. AI becomes useful here not by adding another conversational layer, but by helping remove unwanted layers. The living room has finally found an automation project with an understandable goal: make the screen less interested in everything except the program.

Share articleOpen article permalink →

Nvidia Offers Agents a Place to Stay In

Nvidia has launched new open models aimed at AI agents running inside enterprise data centers and on personal computers. The release follows business interest in operating models on their own infrastructure and in keeping more agent activity on the device. The appeal is control over where computing happens. Local deployment can change an organization's choices about data handling and operations, but it also puts more responsibility on the organization running the system. An agent moving into the company data center has not magically become a well-managed employee. Someone still has to assess permissions, maintenance, and useful performance. The cloud may no longer be the landlord, but the office now gets to discover the joys of owning the building.

Share articleOpen article permalink →

Pritzker Gives AI a Very Wide Performance Review

Illinois Gov. JB Pritzker is weighing AI's potential benefits alongside its risks, discussing technology and privacy while entertaining possibilities as different as longer lives and threats comparable to nuclear technology. Those are expressions of stakes, not established outcomes. For policymakers, that range creates a difficult assignment. Rules must address present decisions without assuming that either the most optimistic promise or the most alarming scenario has already arrived. The political challenge is translating vast possibilities into questions people can evaluate: who has control, what protections apply, and how institutions respond. AI discourse often offers a choice between paradise and catastrophe. Government still has to work in the less cinematic middle, where the agenda includes privacy, accountability, and a meeting that ends on time.

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

Agents Shop, Robots Practice, COBOL Waits Patiently

Stripe reports more agent purchases through Link, while testing, elections, and older programming languages shape AI's next tasks.

Making AI Headlines newspaper, page 10: Agents Shop, Robots Practice, COBOL Waits PatientlyPage 1 of 10
Page 10 of 10. Read the same stories below.

The Shopping Bot Has Entered Checkout

Stripe says AI agents are increasingly using its Link payment service to shop for users, according to Upstarts Media. The company has not shared transaction numbers, leaving the scale of that activity unquantified. The practical question is how delegated shopping preserves the buyer's intent. Finding an item and authorizing payment are different responsibilities. Convenience depends on the boundaries between them being clear. The ideal assistant saves a trip through checkout. The less appealing version merely makes impulse buying happen at machine speed, with excellent punctuation.

Share articleOpen article permalink →

Robots Rehearse Their Bad Days

University of Michigan spinout Laplace Intelligence uses AI agents and interactive simulation to stress-test robots and self-driving vehicles before deployment. The profitable startup's work focuses on finding problems before machines encounter real-world conditions. Simulation offers a place to examine difficult situations without treating public space as the first rehearsal. Its usefulness depends on how well testing exposes meaningful weaknesses. This is a refreshingly practical AI ambition: invent trouble early so fewer surprises arrive later. A robot's most valuable virtual experience may be learning that the world declines to follow its demonstration script.

Share articleOpen article permalink →

Democrats Consider Running Against the Loading Screen

Democrats see a potential political advantage in cracking down on AI as they seek control of Congress in November. If they regain power, they will face choices about how to approach the technology, not simply whether to criticize it. Campaign positioning is easier than writing workable policy. Public concerns can produce momentum, but governing requires deciding which practices need rules and what those rules should accomplish. The electoral pitch may be protection from disruption. The legislative assignment is less glamorous: turn that promise into something more precise than a sternly worded progress bar.

Share articleOpen article permalink →

COBOL Receives a Visit From the Future

OpenAI says it is improving ChatGPT's work in programming languages with fewer resources, introducing evaluations for software engineering, SQLite, and COBOL migration. The effort directs attention toward tasks beyond the most fashionable coding demonstrations. The practical value lies in whether tools can handle specialized work reliably. Success on familiar examples does not automatically transfer to less represented languages or migration projects. Technology loves announcing that everything has changed. Older code has a quieter response: excellent, please preserve the behavior. The future's first assignment may be understanding why the existing system still works.

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 ↑