Omni Topic - Tech

A patch pull request for a path traversal flaw in the OCaml HTTP library cohttp 6.3.0 drew scanning traffic within roughly ten minutes of going public. An AI agent, given only a vague description of the bug type, produced working exploit code in under a minute. The incident raises hard questions about whether traditional security embargo practices still hold up. What Happened A pull request fixing a path traversal vulnerability in cohttp 6.3.0, an HTTP client/server library for OCaml, was published on GitHub. Within about ten minutes, real servers started receiving probe requests matching the exact pattern described in that PR. Separately, a researcher gave an AI agent only a rough description of the vulnerability category — not the actual patch, not a detailed writeup — and the agent produced functioning exploit code in under one minute. No deep familiarity with the codebase or the diff was required. Background: Why Embargoes Used to Work The standard practice in vulnerability ...

OpenAI has announced it will end its agreement to supply AI models to Cursor, the popular AI-powered coding assistant, following its acquisition by SpaceX. The cutoff is scheduled for November 12, 2026, with OpenAI applying the maximum notice period allowed under its contract. Future OpenAI model releases will not be made available to Cursor at all. What Happened OpenAI confirmed it is terminating the contract under which it supplies its models to Cursor, a widely used AI coding tool built on top of large language models. The decision follows SpaceX's acquisition of Cursor. According to OpenAI's own announcement, access will be cut off on November 12, 2026. OpenAI stated it is using the longest notice period permitted under its existing agreement, giving Cursor's developer base time to migrate before support ends. Crucially, the company also made clear that any models it releases going forward will not be offered to Cursor, meaning the relationship isn't just wi...

Orbify has released a web-based navigation demo called Demo 2 v72 that renders terrain as a curved surface instead of a flat plane. The visual effect resembles the folding cityscapes from the film Inception, applied here to real-world-style map data. It runs entirely in the browser, with no app installation required. What Happened Orbify's Demo 2 v72 is a browser-based demo that bends roads and terrain into a curved, sphere-like surface rather than displaying them flat. The result looks like a landscape wrapped around a globe segment, which has drawn comparisons to the bending city scenes in Inception. Controls are straightforward: W, A, S, D move the camera, left-click drag pans the view, and right-click drag rotates it. Anyone familiar with basic 3D game controls can pick it up immediately without a tutorial. Alongside the current version, Orbify also keeps an earlier build, Demo 1 v38, available on the same page, letting visitors compare how the rendering approach has evol...

Google has released Gemini 3.5 Transcribe, a speech-to-text model line built to handle messy real-world audio and still output clean, readable text. The lineup includes gemini-3.5-transcribe-live for bidirectional streaming through the Live API with sub-one-second latency, plus a standard variant aimed at recorded audio. The pitch is straightforward: background noise, technical jargon, stutters, and mid-sentence self-corrections shouldn't end up cluttering the final transcript. What Google Announced Google introduced Gemini 3.5 Transcribe as a pair of speech-to-text models focused on accuracy in noisy, unscripted conditions. According to Google, the models are designed to process background noise, domain-specific terminology, stuttering, and self-corrections mid-sentence, then produce a cleaned-up transcript rather than a literal word-for-word dump of everything spoken. The streaming variant, gemini-3.5-transcribe-live, runs through the Live API and delivers bidirectional stre...

A new analysis of roughly 460,000 GitHub pull requests found a distinct vocabulary cluster resembling Claude's writing style jumped from 0.7% in early 2025 to nearly 39% by mid-2026. The study never looked at the actual code changes, only the words contributors used to describe them. Specific terms like 'load-bearing,' 'seam,' and 'byte-identical' turned out to be reliable markers of this shift. What Happened An independent study clustered roughly 460,000 GitHub pull requests purely by the vocabulary used in PR descriptions and commit messages, ignoring the underlying code entirely. It found that a specific writing-style cluster grew from just 0.7% of PRs in early 2025 to nearly 39% by mid-2026 — a shift large enough to reshape what a 'typical' PR description looks like across a huge, real-world sample. The Telltale Words The cluster is marked by recurring word choices: 'load-bearing,' 'plainly,' 'quietly,'...

A new lawsuit alleges that Elon Musk's xAI trained its Grok models using both real and AI-generated child sexual abuse material (CSAM). The allegations, reported by Ars Technica, raise serious questions about how large AI labs source and vet the massive datasets used to train chatbots and image generators. No final court ruling has been issued, but the claims add to mounting scrutiny of AI training practices industry-wide. What the Lawsuit Claims According to reporting from Ars Technica, a lawsuit filed against xAI alleges that the company's Grok models were trained on datasets containing child sexual abuse material, including both real images and AI-generated CSAM. The suit reportedly surfaced as part of ongoing legal action against xAI. Details about the plaintiffs, the specific evidence cited, and xAI's formal response have not been widely disclosed at this stage. xAI has not issued a detailed public rebuttal covering the specifics of the allegations as of this wri...

A CEO reportedly laid off a developer to free up budget for AI tools. In response, that developer and collaborators released Open Executive, an open-source system that simulates an entire company leadership team. The project raises an uncomfortable question for management: if AI can replace engineers, why not executives too? What Happened The origin story behind Open Executive is almost too on-the-nose. A CEO fired a developer, citing plans to "make room for AI" in the organization. Rather than walking away quietly, the developer built something pointed: an open-source virtual executive suite capable of handling the kind of decisions typically reserved for C-level staff. The project, hosted on GitHub under SenteLabsAI, is called Open Executive. It's not a chatbot wrapper or a single assistant — it's structured to mimic an actual leadership team, complete with department-specific reasoning. How the System Works At the core of Open Executive sits an orchestrator...

A growing number of US restaurants are ditching tipping altogether, folding labor costs directly into menu prices instead. San Francisco's La Cigale is one of the latest examples: a fixed $140 per person, no tip expected, with wine staff paid a guaranteed $40 an hour. The move reflects a longer-running debate in American dining over whether tipping is still a fair way to pay restaurant workers. What's happening At La Cigale in San Francisco, diners pay a flat $140 per person for a set menu and nothing more — no tip line, no percentage math at checkout. The restaurant instead pays its wine staff a fixed $40 hourly wage, regardless of how busy the night is or how generous customers feel. The pitch to diners is simple: the price on the menu is the price you pay. Background: why tipping is under pressure Tipping has long been the backbone of server pay in the US, often making up the majority of a front-of-house worker's income. But it comes with structural problems. Ear...

Perspec 1.0 is a desktop application that automatically straightens the perspective distortion in photos of documents and receipts, released after nine years of development by developer Adrian Sieber. Instead of the edge-detection method most mobile scanner apps use, Perspec segments the photo into document and background regions and then extracts the four corner points directly. Users can manually adjust the detected boundary whenever the automatic detection misses. What Perspec 1.0 Does Developer Adrian Sieber has released version 1.0 of Perspec, a desktop application that straightens the perspective distortion in photos of documents and receipts. Perspec runs as a standalone desktop tool rather than a mobile app, and it lets users manually adjust the detected document boundary whenever the automatic detection gets it wrong. A Different Detection Approach Most mobile scanning apps rely on edge detection: they scan the image for straight lines and infer the document's bounda...

Apple has confirmed that Hide My Email addresses created through iCloud+ will continue to use the icloud.com domain, reversing an earlier plan after developer and user feedback. Starting later this year, however, new private relay addresses generated through Sign in with Apple will be issued under a different domain: private.icloud.com. The two privacy features, which are often confused, are now diverging in how their addresses are structured. What Apple Announced In a brief developer news post, Apple clarified the domain policy for two related but distinct privacy features: Hide My Email (part of iCloud+) and the private email relay used by Sign in with Apple. According to the notice, Hide My Email addresses - the disposable, forwarding email addresses iCloud+ subscribers can generate for sign-ups, newsletters, or online forms - will keep using the icloud.com domain going forward, as they always have. This matters because Apple had apparently considered consolidating or changing ...

Zillow and Redfin have reached a settlement with the Federal Trade Commission over an antitrust case tied to their rental listings arrangement. As part of the deal, Redfin is required to reenter the rental advertising business. The case centered on whether the two companies' prior agreement reduced competition in how rental listings were marketed online. What Happened Zillow and Redfin have settled an antitrust case brought by the FTC. The settlement's central requirement is that Redfin return to the rental advertising business, a market it had stepped back from under a prior arrangement with Zillow. The agreement resolves the FTC's legal action without further litigation, though it imposes a structural remedy rather than just a monetary penalty. Background: The Rental Listings Arrangement Zillow has long been the dominant destination for both home sales and rental listings in the U.S. real estate tech market. At some point, Redfin scaled back its own rental listing...

A humanoid robot has posted a faster 100-meter time than Usain Bolt's famous 9.58-second world record. At a preliminary heat of the World Humanoid Robot Games in Beijing on Saturday, Tiangong Ultra, built by the Beijing Humanoid Robot Innovation Center, completed the sprint in 9.39 seconds. Honor's own humanoid, Lightning, followed closely at 9.47 seconds. What Happened The result came out of a qualifying round, not a final, at an event specifically designed to test humanoid robots across athletic-style categories. Tiangong Ultra's 9.39-second run and Lightning's 9.47-second run both undercut Bolt's 2009 record set at the World Championships in Berlin, a mark that has stood in human athletics for over 15 years. It's worth being precise about what this comparison actually means. These are robots running on a track built for the competition, timed under the games' own rules rather than IAAF regulations. Bipedal robots also move very differently from...

IEEE's next wireless LAN standard, internally called Ultra High Reliability (UHR) and marketed as Wi-Fi 8, is not chasing a higher top speed number. Instead, it targets consistent throughput and lower latency in crowded, interference-heavy environments. That's a notable shift after several generations where headline speed was the main selling point. What's Different About Wi-Fi 8 Most of the raw specifications carried over from Wi-Fi 7 remain unchanged in Wi-Fi 8: maximum theoretical data rates, the number of spatial streams, 4096-QAM modulation, the supported frequency bands, and the 320MHz channel width are all roughly the same. On paper, that makes Wi-Fi 8 look like a minor update. The real change shows up in a different metric. According to figures cited from IEEE documentation, throughput measured across varying signal-to-interference-plus-noise ratio (SINR) conditions improves by around 25% compared to Wi-Fi 7. In plain terms, that means better performance spec...

A new reference site called htmlcat.net catalogs ways to build UI and interactivity using only the HTML, CSS, and JavaScript that browsers already ship with, no external libraries required. It covers CSS features like :has(), @scope, and Container Queries alongside native components like the dialog element and the Popover API. For developers who reach for a library every time they need a modal or a tooltip, the site is a reminder of how much has quietly moved into the browser itself. What the site actually is htmlcat.net is organized around a simple premise: before installing a UI library, check whether the browser can already do it. The site lists concrete techniques, grouped by CSS and by HTML/JS, for building components that used to require third-party code. On the CSS side, it highlights the :has() selector, which lets a parent element be styled based on what it contains, effectively enabling parent-based conditional styling that was previously impossible without JavaScript. I...

Gleam bakes deprecation markers, todo annotations, generics, pattern matching, and error handling directly into the language itself rather than leaving them to libraries or convention. That built-in consistency is what gives Gleam code its notable simplicity and readability. This piece breaks down what that actually means for developers evaluating statically typed functional languages. What Gleam Does Differently Gleam is a statically typed functional language that compiles to Erlang bytecode for the BEAM virtual machine and also targets JavaScript. What's drawing attention right now isn't a flashy new feature but how consistently the language handles a handful of everyday developer concerns: marking deprecated code, flagging unfinished work, generics, pattern matching, and error handling. In most languages, these are bolted on through libraries, linters, or team conventions that drift out of sync with the actual code over time. Gleam instead treats them as first-class la...

A developer named Mat Duggan built a spreadsheet tracking OpenTelemetry's (OTel) issue and pull request history, and the numbers point to an uncomfortable conclusion: the project that succeeded at vendor neutrality is now struggling to ship features at a pace its adopters expect. OTel is the de facto instrumentation standard in cloud-native environments, backed by the Cloud Native Computing Foundation (CNCF). Duggan's analysis argues the project's own process has become its biggest bottleneck. What's Happening A developer named Mat Duggan compiled a spreadsheet tracking OpenTelemetry's issue and PR history across its repositories, and the data tells an uncomfortable story: the project that succeeded at vendor neutrality is now struggling to ship features at the pace its adopters expect. OTel is the de facto standard for instrumentation in cloud-native environments, backed by the CNCF, but Duggan's analysis argues that its own process has become the bott...

Go is a compiled language, which means it has traditionally lacked a native REPL for quick experimentation. gomacro fills that gap by implementing an interpreter for Go, written in Go itself, that runs both as an interactive shell and as a script executor. According to its project page on GitHub, it now also supports Go generics and a macro system, features not found in the standard toolchain. What gomacro Does gomacro (hosted at cosmos72/gomacro on GitHub) describes itself as a nearly complete Go interpreter. It lets developers type Go code line by line into a REPL and see results immediately, without compiling a binary first. It can also execute existing `.go` script files directly. This matters because Go's design philosophy has always favored compiled, statically-typed execution over dynamic scripting. Tools that bring REPL-style interactivity to Go have existed before (such as `yaegi`), but gomacro distinguishes itself by explicitly supporting generics, introduced in Go 1...

Harvard Business School's startup bootcamp, Foundry, now uses AI avatars modeled on its own instructors to coach founders through practice pitches and mock board meetings. The program costs $699 and treats the AI avatars as a supplement to human mentorship rather than a replacement. It's a concrete example of AI moving from chatbot Q&A into structured, repeated skills training. What's happening According to TechCrunch, Harvard Business School's Foundry program has introduced AI avatars built to resemble and sound like its real instructors. Participants pay $699 to join, and during the bootcamp they interact with these avatars while rehearsing startup pitches and simulating board meetings. The avatars aren't simple text chatbots. They're designed to run through realistic scenarios — asking follow-up questions during a pitch, pushing back like a skeptical board member — and then give feedback on delivery and content. Context: why this is happening n...

A growing number of professionals are quietly tuning out documents that read like they came straight from an AI model. A recent essay describes the moment recognition kicks in — a design doc, a marketing deck, a requirements file — and how attention immediately drops. The pattern isn't about AI use itself, but about how obviously unedited the output is. What's Happening An essay circulating among developers and knowledge workers describes a specific kind of fatigue: the moment you spot low-effort AI-generated content in a work document, your focus collapses. Instead of trusting what's already written, you end up going back to the sender and asking them to explain it again — even though the answer is technically right there on the page. The examples given are concrete. A design document laced with phrasing characteristic of Claude. A 20-page marketing deck padded with statements that sound technical but carry no actual information. A requirements document so rambling ...