SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingHacker News ·

OpenAI Used Its Own AI Models to Help Design Its Custom Jalapeño Chip

OpenAI leveraged its own large language models to assist in the design of its custom AI chip, internally codenamed Jalapeño. The company applied LLMs as tools within the chip engineering workflow, marking a notable case of AI being used to build AI hardware. This approach highlights a growing trend of using generative AI to accelerate and optimize semiconductor design processes. The development was reported by IEEE Spectrum, a leading engineering and technology publication.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

FlowScript 0.1 Debuts as a Semantic Language for Modeling Apps Before Coding

Developer Erland Kjensli has publicly released FlowScript 0.1, an experimental domain-specific language designed to describe the semantic model of an application before any implementation begins. The project addresses limitations of traditional flowcharts, which struggle to capture the full complexity of modern apps including state, navigation, UI structure, and dataflow. FlowScript separates structural hierarchy, navigation graphs, and state graphs as distinct but related concepts, aiming to reduce ambiguity in application modeling. A key hypothesis behind the language is that responsive UI layouts can be treated as projections of a single underlying semantic model rather than separate application hierarchies. The initial release is available as a working draft on GitHub under version 0.1.0.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Encoding API contracts in tool schemas beats prose descriptions for smaller LLMs

A developer team discovered in May 2026 that two AI models — Opus 4.7 and Sonnet 4.6 — produced different JSON outputs from the same tool description, with the smaller model returning an incorrect structure that triggered downstream 422 errors. The root cause was not a capability difference but an exposure gap: the larger model had seen the proprietary field names during training, while the smaller one guessed plausible-sounding alternatives. Because tool descriptions are prose rather than enforced contracts, models without prior exposure to a custom schema have no validated structure to follow. The team resolved this by migrating the structural contract into the JSON schema itself, using discriminated oneOf branches with strict required fields for the nine most common pattern types. They applied a permissive fallback branch for rare pattern kinds, limiting the upfront effort to roughly one afternoon while targeting the failures that actually occurred in production.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Microsoft's Record September 2026 Patch Tuesday Fixes 974 CVEs, Two Already Exploited

Microsoft's September 2026 Patch Tuesday addressed 974 CVEs — its largest-ever monthly release — including two zero-day vulnerabilities already being actively exploited at the time of release. One exploited flaw, CVE-2026-81963, is a local privilege escalation in the Windows Update Stack itself, meaning the patch-delivery mechanism became an attack surface capable of elevating a low-privileged user to SYSTEM. The other, CVE-2026-85880, is a heap buffer overflow in Windows ALPC that allows a sandboxed AppContainer process to escape and reach SYSTEM-level privileges. Both flaws require existing local access and serve as escalation steps in broader attack chains rather than standalone entry points. CISA added both vulnerabilities to its Known Exploited Vulnerabilities catalog on 8 September 2026, setting a federal remediation deadline of 22 September 2026.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Track Whether AI Search Engines Like ChatGPT Cite Your Website

As AI-powered answer engines such as ChatGPT, Perplexity, Gemini, and Claude increasingly respond to user queries directly, businesses are finding that traditional SEO metrics no longer capture the full picture of their online visibility. Two emerging disciplines — Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) — focus on whether AI systems select and accurately represent a company's content as a source. Unlike conventional search rankings, an AI citation can mention a brand without linking to it, misrepresent its offerings, or omit it entirely, making structured tracking essential. Marketers are advised to log queries, timestamps, cited URLs, and competitor mentions across multiple AI platforms to detect patterns and measure the impact of content changes over time. Dedicated GEO monitoring tools currently range from around $29 to over $139 per month, with pricing varying based on the number of platforms covered, query frequency, and markets tracked.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Discovers AI Agent Was the Cause of Missing Disk Space

A developer recently shared an ironic experience where an AI agent was tasked with investigating missing disk space on their system. The investigation revealed that the agent itself was the culprit, having consumed the storage it was sent to find. The anecdote quickly gained attention in developer communities for its humorous and self-referential nature. It highlights a growing concern about the unintended resource consumption of autonomous AI agents. The incident has been dubbed a distinctly modern tech problem, reflective of how AI tooling is evolving in 2025 and beyond.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Oracle AI Agent Studio's Retirement Assistant Simplifies Fixed Asset Retirement Process

Oracle AI Agent Studio offers a Retirement Assistant that replaces the traditional multi-step fixed asset retirement workflow in Oracle Fusion Cloud with a conversational interface. The tool allows accountants to retire assets by typing free text or uploading a CSV file, eliminating the need to navigate multiple menus or work areas. The assistant validates inputs in real time, identifying missing fields such as retirement dates and flagging errors with specific, actionable guidance before any changes are posted. It works alongside a companion agent called the Retirement Request Assistant, which lets business users submit retirement requests that are then routed to accountants for processing. Together, the two agents form a front-office and back-office pairing designed to streamline the end-to-end asset retirement workflow.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

US troop deaths in Iran conflict higher than official Pentagon figures, report says

A Washington Post investigation has found that American military deaths during the US-Iran war are undercounted by at least four compared to official Pentagon figures. The discrepancy suggests the Department of Defense has not fully disclosed the true human cost of the conflict. The report raises questions about transparency and accountability in how the military reports casualties. Reuters cited the findings, drawing attention to the gap between official statements and on-the-ground reality.

0 comments Read more at Hacker News
0
ProgrammingHacker News ·

Report Details US Military Kill Chain Behind Strike on Iranian School

A Bloomberg investigation has examined the sequence of US military decisions that led to an airstrike destroying a school in Iran. The report traces the so-called 'kill chain' — the chain of command and targeting process — behind the incident. The strike has drawn significant attention given the civilian nature of the targeted site. The investigation raises questions about the targeting procedures and oversight involved in authorizing the attack.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

SQL Joins Explained: How to Link Database Tables Without Duplicating Data

SQL joins allow developers to combine data from multiple related tables using shared key columns such as customer_id or product_id. The four main join types — INNER, LEFT, RIGHT, and SELF — each serve a distinct purpose depending on whether unmatched rows need to be included. An INNER JOIN returns only rows with matching values in both tables, while a LEFT JOIN retains all rows from the left table even when no match exists on the right. A SELF JOIN enables a table to be compared against itself, useful for finding rows that share a common attribute. INNER JOIN and LEFT JOIN cover most real-world use cases, making RIGHT JOIN largely redundant in practice.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

CISA Adds LiteLLM and Kestra AI Infrastructure Flaws to Exploited Vulnerabilities List

The US Cybersecurity and Infrastructure Security Agency (CISA) added vulnerabilities in LiteLLM and Kestra to its Known Exploited Vulnerabilities catalog on September 2, 2026, marking the first time AI infrastructure components have appeared alongside traditional software flaws on the list. LiteLLM, an AI gateway, carries an improper authentication flaw rated CVSS 8.8, while Kestra, a workflow orchestration platform, has an OS command injection vulnerability rated a critical CVSS 10.0. A ZoomEye internet scan conducted on September 19, 2026, found over 34,000 internet-facing LiteLLM instances and 126 Kestra deployments, highlighting meaningful public exposure. Because these platforms routinely store sensitive credentials such as model provider API keys and database tokens, a successful exploit risks large-scale credential theft rather than mere service disruption. Security researchers are urging organizations to take these AI services offline from public internet access, apply patches immediately, and rotate any potentially exposed credentials.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Machine Learning Explained: How Computers Learn From Data to Make Decisions

Machine Learning (ML) is a branch of Artificial Intelligence that enables computers to identify patterns in data and make predictions without being explicitly programmed for every scenario. It encompasses three main approaches: supervised learning, which trains on labeled data to predict outcomes; unsupervised learning, which discovers hidden groupings in unlabeled data; and reinforcement learning, which improves through trial, error, and reward signals. ML is already widely deployed across industries, with banks using it to flag fraudulent transactions and healthcare providers applying it to assist in diagnostics. Despite involving complex algorithms and mathematical concepts, the core principle remains straightforward — feed data into a system, let it learn patterns, and apply those patterns to generate useful predictions or insights.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds ESP32 Wi-Fi Multi-Tool With Packet Sniffer and Deauth Radar

A developer has detailed the third phase of converting an ESP32-S3 microcontroller into a portable network analysis multi-tool. The project adds a wireless web portal that allows DuckyScript payloads to be uploaded over Wi-Fi to onboard LittleFS flash storage, eliminating the need for a wired serial connection. A packet sniffer running in promiscuous mode captures raw 802.11 frames and saves them as .pcap files directly to memory. The build also includes a Deauth Radar that monitors for Wi-Fi deauthentication and disassociation frames in real time, triggering an on-screen LVGL warning and logging attack details to a file. Streaming file uploads in small chunks and using FreeRTOS queues were key architectural choices to avoid out-of-memory crashes on the RAM-limited device.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Anthropic and Accenture Partner to Scale Claude AI Deployments Into Production

Anthropic and Accenture have announced a multi-year partnership aimed at helping enterprises move beyond AI pilots and deploy Claude-based solutions at scale. A dedicated Accenture Anthropic Business Group will anchor the collaboration, with Accenture planning to train around 30,000 professionals on Claude. The partnership also makes Accenture a premier partner for Claude Code, Anthropic's coding capability, and includes a jointly established Claude Center of Excellence. A co-developed offering will help CIOs measure AI value and accelerate adoption within engineering teams. Initial industry focus will cover financial services, life sciences, healthcare, and the public sector, where regulatory and operational demands are especially complex.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Cloud-Native Portfolio Using Terraform, AWS Lambda, and GitHub Actions

A developer with an IT Operations background completed the Cloud Resume Challenge, hosting a personal portfolio website using cloud-native infrastructure on AWS. Rather than using website builders like Wix or Squarespace, they built the project using Terraform for infrastructure-as-code, which they found more instructive than studying for their Terraform Associate certification. They leveraged an on-premises HashiCorp Vault running on a Kubernetes cluster to manage secrets and JWT authentication, keeping sensitive credentials off the cloud and behind a VPN. CI/CD was handled through GitHub Actions across two repositories — one for the frontend and one for the backend — each with separate testing and production branches. The project also incorporated an AWS Lambda-powered visitor counter with localStorage to prevent duplicate counts on page refresh.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Retrofit Smart Locks Adopt Matter over Thread for Secure, Non-Destructive Access

Retrofit smart locks mount on the interior side of existing door cylinders, allowing tenants and property owners to add digital access control without drilling or structural changes. Unlike early Wi-Fi-dependent locks, newer devices using the Matter over Thread protocol operate on a low-power IPv6 mesh network, reducing battery drain and cloud dependency. The Matter standard enables direct local communication with major platforms including Apple HomeKit, Google Home, and Amazon Alexa, even without internet connectivity. Security is reinforced through Diffie-Hellman key exchange and 256-bit AES encryption on every lock operation, with annual audits conducted by cybersecurity institute AV-TEST. Companies such as Nuki have helped mainstream this retrofit approach across European and Turkish markets, positioning it as a practical standard for apartments and heritage properties.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Why DNS-Level Blocking Is Replacing Browser Ad Blockers for Full Privacy

Traditional browser-based ad blockers like uBlock Origin cannot intercept telemetry from operating systems, mobile apps, or IoT devices, leaving significant privacy gaps. Google's Manifest V3 update further limits browser extensions by capping the number of dynamic blocking rules they can apply. DNS-level filtering using encrypted protocols — DNS-over-HTTPS (DoH) and DNS-over-TLS (DoT) — addresses these gaps by blocking malicious or tracking domains before any network connection is established. Response Policy Zones (RPZ) allow recursive resolvers to sinkhole harmful domains network-wide, covering every device including smart home hardware. Combining encrypted DNS with client-side tools is now considered a more comprehensive approach to privacy than browser extensions alone.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer builds Spotify desktop pet controller using Godot 4 in first engine project

A developer transitioning from Unity created BelPet, a desktop pet application built with Godot 4 and C# that also functions as a Spotify playback controller. The draggable character sits above the Windows taskbar and lets users play, pause, skip tracks, seek, and adjust volume without opening the Spotify app. Because Spotify's Web Playback SDK is browser-only, the developer embedded a WebView inside Godot using the godot_wry library to run the required JavaScript. BelPet also uses a transparent, borderless, always-on-top window with custom mouse passthrough so clicks register on apps beneath it. Spotify's Development Mode restrictions limit full multi-device control for general users, though a workaround exists by creating a personal Spotify developer application.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

BBC Asks Whether Societies Are Psychologically Prepared for Modern War

A BBC article raises concerns about whether populations are mentally and emotionally equipped to face the prospect of armed conflict. The piece examines the psychological readiness of civilians in an era of heightened global tensions. It explores how decades of relative peace in many regions may have left societies unprepared for the realities of war. The article appears to draw on expert perspectives around resilience, fear, and public morale. The discussion reflects growing unease among analysts and policymakers about the shifting security landscape.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Gemma 4 Runs 1.79x Faster on Tesla T4 Using QAT Int4 Weights Over bf16

A developer has published a step-by-step guide for deploying Google's Gemma 4 2B model on a Tesla T4 GPU hosted on a Google Compute Engine VM. Two model checkpoints were benchmarked: the standard bf16 version and a QAT int4-weight variant, using vLLM 0.29.0 on a modest n1-standard-2 instance with 2 vCPUs and 7.8 GB RAM. The QAT model decoded at 72.31 tokens per second in a single stream compared to 40.44 for bf16, and reached 215.91 tokens per second across eight streams versus 164.62. A suite of Python MCP tools was built to manage the deployment pipeline, handling steps such as GPU architecture verification and a required Turing-specific patch. The full setup and source code are available on GitHub, offering a practical reference for running modern LLMs on older, lower-cost GPU hardware.

0 comments Read more at DEV Community
← NewerPage 1136 of 3220Older →