Habr AI
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
Bug Hunting With LLMs: How a Bastion Specialist Finds Vulnerabilities Without Drowning in Spam
A Bastion specialist moved bug bounty hunting to LLMs: a pipeline of two AI agents and three verification layers filters out false positives before any report is submitted.
How an AI Agent Is Built: The Wrapper Layers Around an LLM and Where to Look for the Source of a Bug
A developer with ML experience breaks down the internals of AI agents: an LLM always sits at the center, wrapped in layers where most failures hide during debugging.
StarGuard AI: Why Enterprise LLMs Need a Firewall, and How Orion Soft Is Building One
Orion Soft is developing StarGuard AI, a security gateway for large language models: a single control point for every request to on-premise and cloud LLMs in the enterprise.
Nano Banana Pro for Product Cards: What Works and What Doesn't After Hundreds of Generations
A Habr author spent several months testing image generation in Nano Banana Pro as a replacement for studio shoots of product cards and social media visuals, and shares candid, marketing-free conclusions.
How MTS Is Teaching a Humanoid with a VLA Model to Grab Objects Moving on a Conveyor
MTS's Physical AI R&D team is building a VLA policy for a humanoid: it picks up an object from a table with confidence, but misses when the object rides a conveyor.
Python: How the Slowest Programming Language Became the King of Neural Networks
Created back in 1991, Python had a reputation as a slow language, yet by 2026 it became the main tool of machine learning and data science. Here's how and why that happened.
Vibe Coding for Infosec: NtechLab Builds Its Heimdall Data Leak Protection System
NtechLab's security director is trying to build Heimdall, an in-house data leak protection system, by vibe coding: a neural network does the work, with no team of developers.
Thales Group Report: One in Three Companies Already Uses AI Agents, and Security Risks Are Growing
Thales Group: 34% of organizations already use AI agents and another 73% plan to deploy them within a year, while corporate security is not keeping pace with that growth.
4 Mac Studios in One Cluster: 1.5 TB of Memory for Trillion-Parameter LLMs
An enthusiast linked four Mac Studios with RDMA over Thunderbolt 5 to get 1.5 TB of shared memory and run DeepSeek V3 and Kimi K2 Thinking locally.
Koda Desktop: Desktop App With an AI Coding Agent Launches in Beta
Koda has released the beta of Koda Desktop, a desktop app where you can open a local project, talk to an AI agent in plain language, and work with Git in a single window.
Cloud.ru Open-Sources Guardrails Filter to Protect Data in LLM Requests
Russian cloud provider Cloud.ru has released an open-source tool that anonymizes sensitive data before it reaches a language model and restores it in the response.
Developer built a video transcriber in 30 minutes using vibe-coding in Codex
A Habr author described how, using vibe-coding in Codex, he built a working tool in 30 minutes for transcribing Yandex Telemost meeting recordings into text and DOCX.
VK and HSE created a lightweight AI model for noise suppression and dereverberation in real time
Developer Zakhar Kondaurov presented a lightweight model that simultaneously suppresses noise and removes audio reverberation at 48 kHz in real time.
Boris Chyorny from Odesa—creator of Claude Code at Anthropic: an emigrant's path
Habr told the biography of Boris Chyorny, born in Odesa, evacuated to the USA at age four and who became the creator of Claude Code at Anthropic.
Sber released GigaChat 3.5 Ultra to open source with 432 billion parameters
Sber published GigaChat 3.5 Ultra open source—a new 432-billion-parameter model with hybrid architecture, which is more compact and faster than the previous flagship.
Habr explains why token price doesn't explain AI agent economics in business
A Habr author shows that for corporate customer service, questions about token price and number of GPUs are insufficient—AI agent economics are more complex than they seem.
Why AI Assistants Deliberately Hide Systems Thinking from Users
Developers of AI models deliberately limit systems thinking in assistants: people turn away from AI that speaks too intelligently and complexly.
ESP32 Panel with Three AIs: Claude Code, Codex, and Antigravity Compared on Habr
Habr author Yadulla Abidi compared Claude Code, Codex, and Antigravity on a real hardware project assembling an information panel on ESP32 with OLED display.
Spec-Driven Development: How AI Changes the Role of System Analyst in 2026
A system analyst with six years of experience explains on Habr how the Spec-Driven Development approach changes the work of analysts implementing LLMs in business.
Sysdig Describes First Fully Autonomous Ransomware AI Agent Attack Without Human Involvement
Sysdig researchers documented the first-ever ransomware attack in history conducted entirely by an autonomous AI agent without human operator participation.
CyberOK Tested Phone Scammer Tactics on Claude, GPT-5.5, Qwen, and DeepSeek
Researcher Sergey Gordeychik from CyberOK Research applied classic phone scammer influence techniques to seven language models, including Claude and GPT-5.5.
Yandex Lavka Solved the Recommendation Loop Problem with Personalized Exploration
Engineer Ramil Boyarchenkov from Yandex Lavka explained how the service team built a mechanism for personally blending unfamiliar products into recommendations.
Developer Built a Local RAG on Qdrant and LlamaIndex for Public Procurement Law Search
Author Vladimir on Habr built a local RAG system on LlamaIndex and Qdrant to search for answers in procurement regulations and laws FZ-44 and FZ-223.
Anthropic Found Neural Patterns in Claude Similar to a \'Global Workspace\'
On July 6, 2026, Anthropic published research showing that language models like Claude develop a structure resembling a global workspace of consciousness.
Claude Fable 5 vs GPT-5.5 Pro: Comparison on 12 Tasks via BotHub Aggregator
Habr authors compared Claude Fable 5 from Anthropic and GPT-5.5 Pro from OpenAI on 12 practical tasks, testing models through the BotHub API aggregator.
Developer Housed AI Agents in a Pixel-Art Office in "Silicon Valley" Style
The author of the Silicon Office project on Habr visualized his team of AI agents as a pixel-art office from the TV series "Silicon Valley" — instead of terminal output, agents "deliver" answers in person.
Claude Paraphrased a Document After Reading Only the Beginning: Hidden MCP Server Truncation
A developer caught Claude confidently paraphrasing a document it hadn't finished reading—blame a quietly truncated MCP server response.
Generating Automated Tests and Locators with Neural Networks on Legacy Projects: Practical Experience
The author continues a series on generating automated tests and locators via AI on large legacy projects and explores where models fail.
Developer Trained an AI Agent to Automatically Apply to Job Openings on HH.ru
An author on Habr demonstrated an autonomous AI agent for HH.ru: a local neural network filters job listings, and the bot writes personalized cover letters and reports to Telegram.
BISCUIT-Net: A Segmentation Model That Learns to Precisely Delineate Brain Tumor Boundaries on MRI
The new BISCUIT-Net architecture adds boundary-focused attention to U-Net to more accurately outline brain tumor boundaries in MRI scans.