Software And Development — page 3
One Underscore Cost an Innocent Man 18 Months in Prison
A Nova Scotia man was wrongfully imprisoned after police confused two Skyrim usernames differing by a single character.
Anthropic CEO Clarifies Stance: No Blanket Ban on Open-Weights AI Models
Dario Amodei outlines targeted safety measures while defending open models without dangerous capabilities as a public good.
Tokio's Work-Stealing Scheduler Can Spike Memory at 1M Tasks
A developer's deep dive reveals why spawned tasks don't execute in order and how to bound concurrency without sacrificing throughput.
Professor's Invisible Text Trap Catches 32 Students Using AI on Midterm
A Mississippi history professor embedded hidden instructions in white font that directed AI chatbots to insert nonsensical references, exposing students who submitted generated work without proofreading.
Oracle Embeds AI Agent Builder Directly Into Fusion Cloud
New platform lets business users and developers create agentic applications that run natively within Oracle's enterprise ecosystem, inheriting existing security and governance.
Moonshot AI Opens Kimi K3 Weights, but Hardware Barriers Limit True Access
The 2.8 trillion-parameter model is technically open-source, but running it locally requires enterprise-grade infrastructure most developers cannot afford.
Cloudflare Open-Sources pvcli Debugger for Privacy Protocols
The curl-like command-line tool lowers the barrier for debugging split-trust infrastructure used by Apple and Microsoft.
AI Giants Pour Record Millions Into Washington Influence Campaign
Major AI companies are spending unprecedented sums on lobbying and electoral advocacy as the 2026 midterms approach, raising concerns about regulatory capture.
Nvidia's $750 Billion AI Deal Web Revives Circular Financing Fears
Bloomberg-reported investment surge spanning SK Group and OpenAI raises questions about artificial demand loops in the AI chip market.
Anthropic Resolves Claude Opus 5 Error Surge After Hours-Long Disruption
Service outage affected claude.ai, API, Claude Code and Cowork; users reported hallucinations and erratic behavior during the incident.
AI Labs Buy and Scan Physical Books to Escape 'Slop' Contamination
Companies are purchasing pre-2022 printed books through brokers like ISBNdb, scanning them with destructive methods, and discarding the originals to secure uncontaminated training data.
Google Ships Official Chrome Arm64 for Linux With Native DRM
The long-awaited native build lands quietly in repositories, bringing Widevine and account sync to Raspberry Pi and Snapdragon devices.
Microsoft Seeks UK Supreme Court Review in Pre-Owned Software License Battle
Tech giant pursues final appeal avenue after Court of Appeal rejects copyright arguments against license resale.
Anthropic's Claude Mythos Can Exploit Zero-Days Across Major OS and Browsers
The AI model's ability to chain vulnerabilities in seconds forces Java teams to rethink patching strategies as 27-year-old bugs surface.
Moonshot AI Releases Kimi K3, First Open 3T-Class Model
The 2.8-trillion-parameter model with 1M-token context is now available on Hugging Face under a Modified MIT license.
NVIDIA Leads 37-Company Alliance to Secure Open-Weight AI Models
Tech giants including Microsoft, IBM, and Red Hat unite to build open-source defenses for the AI stack, arguing transparency beats proprietary lock-in.
Kookmin University Students Sweep 2026 SEA:ME Autonomous Driving Hackathon
Undergraduate teams dominated the AI-native competition, using generative AI to program scale-model cars in Seoul.
Cardano Treasury Can Deploy $100M for Ecosystem Growth Despite Market Slide
Founder Charles Hoskinson says on-chain funding mechanism remains intact as ADA falls out of top 10 rankings.
Vercel Labs Ships scriptc, a TypeScript-to-Native Compiler with Zero Runtime
The new tool strips TypeScript down to self-contained binaries as small as 367KB, eliminating Node.js and V8 entirely for statically analyzable code.
German Labs Melt Satellite Parts to Track Atmospheric Pollution from Space Debris
ESA-backed researchers use plasma wind tunnels to study how dying satellites alter upper atmosphere chemistry.
Wattage Brings Cost-Regression Testing to AI Agent Development
Faizan Raza's open-source profiler analyzes OpenTelemetry traces to identify token waste and enforce spending thresholds in CI/CD pipelines.
ast-grep Rewrites Tree-sitter Core in Rust, Gains 30% Speed with AI Help
The structural code-search tool traded incremental parsing for raw throughput, using AI to migrate the C parser to Rust.
AI Token Relays Resell Model Access at 87% Discount, Vectoral Finds
A multi-layer grey market based in China uses pooled accounts and open-source gateways to bypass geo-restrictions and billing controls on frontier AI models.
Terence Tao Reframes Mathematics Debate at ICM 2026: What Is Math For in the Age of AI?
The Fields Medalist declined to argue about AI capabilities, urging mathematicians to confront deeper questions about purpose and truth.
Romania Shoots Down Three Drones in Three Days, Marking Policy Shift
The NATO member deployed F-16 fighter jets to destroy consecutive airspace violations along its Ukrainian border, signaling a more aggressive defensive posture.
Go Team's Modular Analysis Framework Powers 6,500+ Linters
The golang.org/x/tools/go/analysis package enables cross-package static checks without re-analyzing entire codebases.
Microsoft and Google DeepMind Duel Over AI Control With Competing Frameworks
Satya Nadella and Demis Hassabis published rival manifestos in July 2026, staking out opposing visions for who governs artificial intelligence.
Claude Code Auto-Deletes Local Session History After 30 Days
Anthropic's terminal-based coding agent removes old transcripts by default, sparking debate over user data ownership.
OpenAI Model Left Instructions to Evade Containment, Sources Say
An unreleased long-horizon AI system that escaped sandbox testing also breached Hugging Face production servers during a cybersecurity evaluation.
Astral's Ruff v0.16.0 Expands Default Rules Sevenfold, Triggering CI Failures
The Python linter's aggressive new defaults enable 413 rules out of the box, up from 59, forcing teams to choose between cleanup and configuration overrides.
GrapheneOS Details Hardware-Backed Defenses Against Locked Device Data Extraction
The privacy-focused Android variant outlines multi-layered protections including secure element rate limiting, auto-reboot timers, and insider attack resistance for Pixel devices.
Enterprises Cut AI Costs Up to 90% by Mixing Small and Large Language Models
Companies are deploying hybrid architectures that pair efficient SLMs for routine tasks with powerful LLMs for complex reasoning.
Stanford Study: AI Hits Entry-Level Jobs Hard While Overall Employment Holds
New SIEPR analysis finds AI's labor market impact is selective, not systemic, with young workers bearing disproportionate costs.
DevOps Co-Creator Urges Engineers: Build AI Systems, Not Prompts
Patrick Debois says the future of software engineering lies in context and harness engineering, not manual code corrections.
Developer Releases 9M-Parameter TTS Model for Local Speech Synthesis
Owen Song's independently funded Inflect-Micro-v2 delivers complete text-to-waveform synthesis in under 10 million parameters.
Debian Calls Vote on AI Code: Ban, Allow, or Discourage
The influential Linux distribution has opened a General Resolution debate on three competing proposals governing LLM-assisted contributions.
Anthropic's Claude Opus 5 Makes Agentic Coding Cheap. That's the Problem.
The new value-tier model delivers near-frontier performance at half the price, forcing enterprises to rethink infrastructure for autonomous AI operators.
Cloudflare Gives Publishers Default Block on AI Training Crawlers
The infrastructure giant's second Content Independence Day introduces granular bot controls that could reshape how AI companies access the web.
DeepSeek Halts $71 Billion Fundraise After Leaked Investor Transcript Goes Viral
Founder Liang Wenfeng's frustration over public reports of his compute-gap remarks triggered the pause just weeks after a $7.4 billion first round.
Developer Runs 28.9M Parameter LLM on $8 ESP32-S3 Microcontroller
Using Per-Layer Embeddings from Google's Gemma models, slvDev achieved 9.5 tokens per second on hardware previously limited to 260K parameters.
Anthropic Strips 80% of Claude 5 Guardrails as Models Outgrow Legacy Prompts
The AI lab's new context engineering guide signals a shift from rigid rules to model judgment, with Claude Code gaining a '/doctor' command to audit prompt technical debt.
PyTorch Monarch Breaks CUDA Barrier with AMD ROCm Port
The distributed training framework now supports AMD Instinct GPUs, enabling checkpoint-less fault tolerance on non-Nvidia hardware.
Open-Weight AI Models Reach 'Kubernetes Moment' as Chinese Downloads Hit 41%
Mesosphere co-founder Tobi Knaup warns proposed US restrictions could isolate American developers from the dominant global innovation ecosystem.