April 2026
Google Presented Auto-Diagnose — an AI System for Finding Causes of Integration Test Failures
Google presented Auto-Diagnose — a system based on Gemini 2.5 Flash that automatically analyzes logs from failed integration tests and ident
OpenAI loses three more executives in a week amid reorganization ahead of IPO
OpenAI announced the departure of Vice President of Science Kevin Vail, head of Sora Bill Peebles, and corporate division executive Srinivas
Selectel: AI doesn't take away jobs, but makes entering the profession significantly harder
Selectel describes a new labor market around AI: vacancies don't disappear, but it's harder for juniors to enter the profession, and demand
OpenAI GPT-OSS: Launching Open-Weight Models in Colab with MXFP4 and Advanced Inference
A new guide explains how to launch openai/gpt-oss-20b in Google Colab: install dependencies for Transformers, check GPU, enable MXFP4 quanti
Habr AI releases guide to ChatGPT, Claude, and mcp for newcomers
Habr AI explains without unnecessary theory the differences between local and cloud models, why to pay for subscriptions, how agents work, a
Three OpenAI executives depart amid Sora shutdown and wind-down of side projects
In a single day, OpenAI lost its former product director, Sora head, and enterprise CTO as the company cuts non-core initiatives and intensi
Anthropic Chief Discussed Mythos Access at White House Following Pentagon Conflict
Anthropic CEO Dario Amodei met with White House leadership to discuss U.S. government agencies' access to Mythos—an AI model for detecting c
Cursor preparing $2B funding round at $50B valuation amid AI-coding boom
Cursor is discussing raising at least $2B at a $50B valuation, nearly doubling its end-2025 level amid explosive demand for AI coding tools.
Joshua Bengio and LawZero: why fear of future AI distracts from today's threats
A text on 'Pascal's wager' in AI explains why fear of hypothetical superintelligence diverts attention from already real threats: surveillan
Anthropic and Claude Opus 4.7: Actual Token Consumption Exceeded Claimed Figures
An author measured Claude Opus 4.7's new tokenizer and found consumption increased up to 45–47%, contrary to Anthropic's promised 0–35%, dir
Palantir, Thales and Air Space Intelligence compete for FAA AI contract for airspace management
FAA develops SMART system designed to predict airspace conflicts two hours in advance, with Palantir, Thales and Air Space Intelligence comp
China: Washington's export bills could disrupt chip supplies
Beijing criticized semiconductor export restrictions being advanced in the U.S. Congress, warning of risks to global trade and stability of
Sony AI: Robot Ace Defeats Elite Table Tennis Players by Official Rules
Sony AI's Ace system won three out of five matches against elite opponents by official rules, demonstrating that robots can already compete
US Seeks to Restrict China's Access to Chip Manufacturing Equipment via MATCH Act
China warned that the new US legislative proposal MATCH Act on export restrictions for chip manufacturing equipment will impact not only Bei
Intel shares hit new record after strong forecast amid AI boom
Intel shares reached a historic high following a revenue forecast that exceeded expectations: the market saw the first clear signs that the
Oakland Court to Hear Elon Musk's Lawsuit Against OpenAI and Sam Altman Over Mission
Federal court proceedings begin in Oakland to determine whether OpenAI violated its nonprofit mission after becoming one of the most expensi
OpenAI releases GPT-Rosalind for biology: capabilities and limits of the new model
OpenAI launched GPT-Rosalind for life sciences tasks and integrated it with a Codex research module, promising to accelerate hypotheses and
Anthropic Restricts Access to Claude Mythos Over Global Cybersecurity Risks
Anthropic is not publicly releasing Claude Mythos: the model can discover and potentially exploit thousands of vulnerabilities, and reports
Nvidia: Jensen Huang calls DeepSeek's launch on Huawei chips a threat to the USA
Nvidia's head warned that DeepSeek's shift from CUDA to Huawei's Ascend and CANN stack could weaken US technological influence in AI, even t
Cursor Security Audit Discovers Four Vulnerabilities in Code Editor Protection, but Authorization Remains Secure
Technical audit of Cursor revealed prototype pollution, a hidden dev field, and internal architecture leaks, but confirmed that subscription
Anthropic Unveiled Claude Mythos: AI for Zero-Day Discovery Shifts Internet Power Balance
Anthropic refuses to release Claude Mythos publicly: the model can find and exploit zero-day vulnerabilities while strengthening defenses an
Manitoba prepares strict ban on social media and AI chatbots for minors
Canadian Manitoba authorities want to block children and teenagers from accessing social media and AI chatbots, shifting the conversation ab
Anthropic released Opus 4.7, and OpenAI turned Codex into a computer work agent
This week in AI brought several shifts: Anthropic updated Opus 4.7, OpenAI gave Codex computer control, and Google and Baidu unveiled new vo
UCL: Hybrid quantum computing and AI predicts chaotic systems more accurately
UCL team showed that quantum computers can find hidden stable patterns in chaotic data, making AI predictions more accurate, stable, and vas
ChatGPT Nailed the Diagnosis in Five Cases, But Failed on Treatment Planning
In a five-case medical comparison, ChatGPT never misdiagnosed the primary condition—but fell noticeably short on practical recommendations,
Beauty Brands Accelerate AI Adoption: From Virtual Selection to Formula Development
Artificial intelligence is increasingly integrated into the beauty industry: it helps customers select products while enabling companies to
Why LLM Services Ignore Your Instructions and How to Actually Regain Control
Even a detailed prompt doesn't guarantee a compliant response: this article explains why LLMs break format, succumb to injections, and requi
Anthropic Predicts Engineering Professions' Disappearance, Hires 429 Developers
Anthropic's head claimed that AI is gradually taking over programming, but the company simultaneously opened hundreds of engineering positio
Anthropic releases Claude Opus 4.7 for agentic programming, vision, and autonomous tasks
Anthropic introduced Claude Opus 4.7 — an update to its flagship model with notable improvements in agentic coding, detailed image analysis,
PrismML Bonsai: How to Run a 1-Bit Model on CUDA with GGUF, JSON and RAG
In a new PrismML Bonsai tutorial, we covered how to run Bonsai-1.7B on GPU via CUDA and GGUF, measure throughput, configure chat, strict JSO
xAI launches separate Grok APIs for speech recognition and synthesis for corporate developers
xAI has split Grok's voice stack into standalone APIs: Speech-to-Text and Text-to-Speech for business with aggressive pricing, multilingual
NVIDIA released Ising — the first open family of AI models for quantum-classical systems
NVIDIA presented Ising — an open family of AI models for quantum processor calibration and error correction to bring useful hybrid quantum-c
Google and OpenAI Hit the Limit: What Happens When the Internet Runs Out of Human Text
Generative AI not only drains website traffic through AI summaries, but also undermines its own training data foundation: the less incentive
Google enables Gemini to tap Gmail and Photos for more accurate, personalized answers
Google's Personal Intelligence feature in Gemini connects Gmail, Photos and other Google services, allowing the AI to understand context wit
Why LoRA Breaks in Production and How RS-LoRA Saves Model Fine-tuning
LoRA excels at style and format, but loses signal when fine-tuning with new facts; RS-LoRA solves the problem by changing the scaling formul
Anthropic Investigates Leak of Protected Mythos Cybersecurity Model to Private Forum
Anthropic is investigating a possible leak of Mythos — a closed AI model for cybersecurity — which allegedly appeared on a private forum and
OpenKB and OpenRouter show how to build a local AI knowledge base with Llama search
A new tutorial explains how to deploy a local knowledge base on OpenKB, connect an open model via OpenRouter, and safely configure search wi
How Moscow Credit Bank Shows the Evolution of Employee Training in Banking — From Clerks to AI
Moscow Credit Bank traced how banks trained employees from coin verification and business correspondence to personalized onboarding, complia
OpenAI integrates Sky technologies into Codex for Mac and enhances background app management
OpenAI has integrated Sky capabilities into Codex for Mac: the agent now manages multiple applications in the background, uses independent c
Strike on SABIC hits AI supply chain: PCB resin prices surge, lead times extend
After a strike on SABIC's facility in Al-Jubail, production of resin for PCB laminates has halted, prices have surged 40%, and delivery time
Rio Tinto and the copper race: how the AI boom forces the US to urgently scale mining
Growing data centers and power grid demand turn copper into a strategic resource: the US is trying to launch new mines, but faces import dep
Google discusses AI chips for inference with Marvell, reducing Broadcom dependence
Google is in talks with Marvell about two specialized AI chips for inference and memory, expanding its silicon partners beyond Broadcom and
Elon Musk and Sam Altman Head to Court: OpenAI Mission Dispute Reaches New Heights
A lawsuit by Elon Musk against Sam Altman and OpenAI begins in California: the dispute over the company's nonprofit mission has become a maj
Habr AI Explained How Data-Driven Solution Search Changes Client-Server Architecture
The Habr AI article examines how expert systems with data-driven solution search should structure user interaction locally, on a local netwo
Seedance 2.0 draws attention with hyper-realistic generation of people and scenes
Seedance 2.0 multimodal system accepts text, images, audio and video, outputting nearly cinematic videos that have already sparked debate ov
Anthropic and White House Discuss AI Mythos After Discovery of Thousands of Vulnerabilities
Anthropic's new model for finding code bugs has identified thousands of vulnerabilities in popular operating systems and browsers, according
Atech from Copenhagen Raises Undisclosed Pre-Seed from Lovable, Sequoia Scout and a16z Scout
Danish startup Atech, which transforms text descriptions of devices into working prototypes, has closed a pre-seed round backed by Lovable,
Claude Code and Codex Are Changing the Rules: Prompts Give Way to Context Engineering
An analysis of Context Engineering reveals why in Claude Code, Codex, and agentic IDEs, it's not the prompt that matters, but how memory, sy
Attack on OpenAI CEO's home escalates fears that AI protest may turn violent
Following an attack on Sam Altman's home and an attempted arson at OpenAI headquarters, the debate over AI risks has shifted from policy and
Why diffusion models leave seams on 40-megapixel photos and how to give tiles memory
The article examines why tiling breaks 40–150-megapixel frames in diffusion models and what memory mechanisms from video diffusion can be tr
China Forced Meta to Cancel $2B Manus Deal and Tightened Rules for AI Transactions
Chinese authorities ordered Meta to cancel its Manus acquisition for $2 billion, setting a precedent that extends control not only to assets
OpenAI and Magika showed how to build a pipeline for file recognition and threat analysis
The guide demonstrates a practical pipeline where Magika determines the actual file type from bytes, and OpenAI helps interpret the result a
MIT researchers teach AI to honestly say "I'm not sure" and hallucinate less
MIT researchers proposed a training method that forces language models to more accurately assess their own confidence and deliver fewer conf
Lubomir Gorbatko Introduces Sessa — An Alternative to Transformers and Mamba
Habr published an analysis of the Sessa architecture, where the author attempts to combine the strengths of Transformer and Mamba to better
MIT Presented EnergAIzer — A Fast Way to Assess AI Energy Consumption in Data Centers
Researchers from MIT and the MIT-IBM Watson AI Lab created EnergAIzer — a tool that assesses AI workload energy consumption in seconds and h
DeepSeek cuts DeepSeek-V4-Pro prices by 75%, reduces repeated requests cost by 10x
DeepSeek has aggressively reduced access costs to DeepSeek-V4-Pro and slashed the price of repeated requests by 10 times, intensifying price
MIT Opens MathNet — World's Largest Collection of Math Olympiad Problems
MIT researchers compiled an open archive of over 30,000 math olympiad problems and solutions from 47 countries to more precisely test AI and
Harvard Business Review: how active AI use leads to "brain overheating"
Harvard Business Review describes the AI brain fry effect: the more actively employees rely on Claude, Gemini, and agents, the higher the ri
Flag Soft: Dali Trial benchmark helped select LLMs by quality, speed, and cost
The Dali Trial benchmark compared LLMs across three criteria important for product implementation: answer quality, processing speed, and tot
UK authorities underestimated AI data center emissions by more than 100 times
The UK government has revised its assessment of AI data center emissions: instead of a barely noticeable effect, the discussion is now about