April 2026
Habr AI reveals how a single open port in an LLM platform exposed API keys and infrastructure
A security audit team testing a RAG service with tens of thousands of users didn't break through complex attack chains, but obtained access
Meta Accelerates Energy Infrastructure: Entergy Raises Investment Plan by $14B Due to Data Centers
American Entergy has increased its four-year capital investment plan to $57 billion: the main reason is gas-fired power plants for a major M
Anthropic Claude Won't Save You From Bad Requirements: Why AI Delivers Convincing But Weak Results
Using Claude as an example, the author demonstrates that AI often accelerates the delivery of plausible answers rather than useful results w
DeepSeek, Google, and Meta: 10 Techniques for LLM KV-Cache Compression to Reduce Inference Memory
KV-cache has become a memory hog for large LLMs, and a new survey reveals 10 approaches — from H2O and SnapKV to TurboQuant and DeepSeek's M
Microsoft exceeded Azure forecasts, but investors worry about AI monetization pace
Azure showed quarterly growth slightly above analyst expectations, but this didn't reassure investors: the market doubts Microsoft is moneti
Elon Musk told the court that OpenAI abandoned its public mission for profit
Elon Musk told the jury that he lost trust in other OpenAI leaders when the project, conceived as a public good, according to his account, s
Republicans in U.S. House of Representatives Investigate Airbnb and Anysphere Over Chinese AI Models
Republicans in the U.S. House of Representatives are investigating whether Airbnb and Anysphere use Chinese AI models, amid Congress's harde
Alphabet Exceeded Expectations Driven by Google Cloud and Enterprise Demand for AI
Alphabet beat expectations: Google Cloud growth and demand from AI clients showed that the company's major infrastructure investments are be
MIT Presents WRING — A Method to Reduce Bias in AI Vision Models Without New Distortions
Researchers from MIT, WPI, and Google created WRING — a method for vision-language models that reduces target bias without amplifying other
Meta shares fall after raising 2026 AI spending forecast to $145 billion
Meta raised its 2026 capital expenditure forecast to $125–145 billion, citing higher component costs and expanded data centers, but investor
Mercor Reaches $10 Billion Valuation by Hiring Specialists to Train AI on Office Work
San Francisco-based startup Mercor, valued at $10 billion, pays qualified specialists to train AI on their own work tasks, betting on automa
Meta tumbles after raising 2026 AI and data center capex forecast
Meta raised annual capital expenditure guidance to $125–145 billion, sparking a selloff as investors worry massive AI investments won't deli
PwC and AWS Demonstrate AI System for Contract Analysis with 90% Verification Reduction
PwC unveiled the AIDA system on AWS, which uses Amazon Bedrock to extract terms from contracts, answer questions about them, and, according
Fresh Apple and Nvidia earnings reveal where the AI infrastructure market is heading
A new wave of quarterly reports from Apple, Broadcom, and Nvidia shows that the AI market is maturing: investors are looking not for promise
Anthropic discussing new round with valuation above $900 billion and may surpass OpenAI
Anthropic is considering a new investment round with a valuation above $900 billion — if the deal closes, the company could become the most
SoftBank Prepares US IPO for AI Robotics Company Roze, Betting on Robots and Data Centers
According to FT, SoftBank is creating a separate Roze company in the US for AI robotics and data centers, aiming to take it public in 2026 w
Former Neural Band leader at Meta seeks funding for Flourish at $2.5B valuation
Thomas Reardon, who led Neural Band at Meta, is raising investments for new startup Flourish, which develops energy-efficient AI and targets
OpenAI achieves key US AI capacity milestone ahead of schedule, accelerating data center expansion
OpenAI reached a critical US AI computing capacity milestone several years ahead of plan, reducing infrastructure risk and expediting data c
Proton CEO: privacy in the age of AI is possible, but autonomous agents pose a new risk
Proton CEO Andy Yen believes that privacy in the age of AI can be preserved through local models and encryption, but he considers autonomous
Gothenburg handed school place distribution to an algorithm — and got chaos with no one accountable
Gothenburg authorities implemented an algorithm for distributing school places, but instead of the promised objectivity, they got chaos, par
White House Opposed Anthropic's Plan to Expand Access to Mythos Model
According to WSJ, the US Administration did not support Anthropic's plan to expand access to the Mythos model, which the company itself cons
Mphasis CEO Nitin Rakesh Explains How the Company Is Reimagining Business Through AI
The head of Indian IT firm Mphasis told Bloomberg why AI is transforming the outsourcing business model and how a mid-sized company competes
Ara Darzi: AI could accelerate fight against antibiotic resistance, but market obstacles stand in the way
British surgeon Ara Darzi believes AI can reduce waiting times during dangerous infections and accelerate the search for new antibiotics, bu
ChatGPT Growth Slows: OpenAI Faces Rising App Deletions and IPO Risks
Sensor Tower records growing ChatGPT deletions and notable cooling of user growth—particularly problematic for OpenAI as the market discusse
Canonical has no plans for a global AI kill switch in Ubuntu, and Linux users debate the move
Canonical's plans to embed AI features in Ubuntu have sparked controversy: users demand an AI-free mode and a 'kill switch,' but the company
Google adds new Gemini features to Google TV with photo, video and Shorts generation
Google has expanded Gemini in Google TV: televisions will get image generation through Nano Banana, video generation through Veo, voice sear
SenseTime Opens SenseNova U1 Model for Chinese Chips, Betting on Speed
Chinese SenseTime released an open multimodal SenseNova U1 model that generates and understands images faster and is already optimized for l
Google announces record number of Search queries and 19% revenue growth
Alphabet reported Q1 2026 results: Google Search set a record for queries, and AI features drove search business growth of 19%
Runway bets on world models, sees AI video as just the first stage
Runway's head says video generation has already become a working tool, but the main market lies ahead: world models could take the company i
Parag Agrawal's Parallel Web Systems Startup Reaches $2 Billion Valuation After New Funding Round
Parallel Web Systems, the startup of former Twitter head Parag Agrawal, raised an additional $100 million from Sequoia and reached a $2 bill
Musk v. Altman Lawsuit Reveals OpenAI Letters, Documents, and Early Architecture
In Elon Musk's lawsuit against Sam Altman, court filings have begun disclosing early details about OpenAI's roles, Nvidia support, and dispu
Google Cloud exceeded $20 billion in quarterly revenue for the first time, but faces capacity shortage
Google Cloud earned over $20 billion in quarterly revenue for the first time thanks to AI, but Alphabet acknowledged that growth is constrai
Taylor Swift Protects Her Image and Voice Against Wave of TikTok Deepfakes in Scam Ads
An investigation revealed how TikTok is spreading AI videos featuring Taylor Swift, Rihanna, and other celebrities to extract personal data
US Emergency Services Complain That Waymo Robotaxis Increasingly Obstruct Emergency Response
American firefighters, police officers, and paramedics say Waymo autonomous vehicles increasingly freeze on roads, block exits, and slow eme
OpenAI Accelerates Stargate and Expands Data Center Network Amid Growing AI Demand
OpenAI reported that the Stargate project has already exceeded its 10 GW target for AI computing capacity in the USA ahead of schedule and i
SoftBank Launches Roze AI: Robots for Data Center Construction and $100 Billion Valuation
SoftBank, according to FT and WSJ, is creating a US-based company called Roze AI that will accelerate data center construction using autonom
OpenAI Explains the Origin of "Goblins" in GPT-5: How a Personality Bug Made It Into the Model
OpenAI revealed why GPT-5.1, GPT-5.4, and GPT-5.5 started more frequently mentioning goblins and gremlins: the cause traced back to training
Shapes: an app where AI characters become group chat participants on equal terms with people
Shapes adds AI characters directly to group chats — they see the entire conversation, react to other participants, and remember the conversa
Eka robots sort nuggets and screw in light bulbs — a ChatGPT moment for physical AI is coming
Eka created robots with eerily realistic movements for delicate manual tasks. Wired asks: is this true physical intelligence or skillful imi
Oracle bets on AI and data centers for OpenAI — market tests whether the boom is deflating
Oracle is increasingly diving deeper into AI infrastructure and building a business around OpenAI's demand, becoming a public indicator of h
OpenAI released a cybersecurity plan in the AI era for government structures and critical infrastructure
OpenAI presented a five-pronged plan: to expand defenders' access to powerful AI tools, strengthen control over their use, and add new secur
How a Russian Developer Spent Days Launching Gemini—and What Finally Worked
A Russian programmer spent several days launching Gemini from Russia via VPN, tried dozens of approaches, and documented each step—what didn
Elon Musk accused Sam Altman in court of 'stealing' nonprofit OpenAI
Elon Musk stated in court that Sam Altman led OpenAI away from its nonprofit mission, and the outcome of the case could now affect both the
Canonical to embed local AI tools in Ubuntu 26.10 with one-click disable
Canonical will add local AI tools based on LLM to Ubuntu 26.10: they will be optional, fully removable, and by default will not send data to
Product Graph and Agent Memory: Why AI Doesn't Save Products Without Knowledge Structure
An analysis of Product Graph explains why even powerful AI agents are useless without shared product memory and how interconnected knowledge
OpenAI prepares to grow its audience 37-fold with ChatGPT Go plan at $8
ChatGPT Plus is forecast to lose 80% of subscribers in 2026 — but the new ChatGPT Go plan at $8 per month is expected to expand the user bas
Poolside released Laguna XS.2 and M.1 — open models for agentic programming
Poolside unveiled two Laguna models for agentic coding: the open XS.2 runs locally, while the more powerful M.1 is designed for long tasks w
LlamaIndex ParseBench: How to Test Document Parsing via Python and Hugging Face
A practical walkthrough shows how to build a document parser evaluation pipeline using the LlamaIndex ParseBench dataset: load PDFs from Hug
OpenAI Explained How ChatGPT Detects Threats and Blocks Dangerous Violence Scenarios
OpenAI described ChatGPT's security system: model-level restrictions, automatic threat detection, manual review of disputed cases, and escal
smol-audio from Deep-unlearning: A collection of Colab notebooks for audio model fine-tuning
Deep-unlearning released smol-audio — a collection of Colab-compatible notebooks for fine-tuning Whisper, Parakeet, Voxtral, Granite Speech
Laboratory Over Six Years: From USB Drives and Notebooks to AI That Finds Hidden Equipment Defects
A story about how one laboratory spent six years building digital infrastructure — and ultimately developed AI that reads deposition logs an
Bank of England warns that rapid AI growth in finance risks systemic failures
Bank of England stated that rapid AI adoption by banks and funds could pose a threat to financial stability, especially if failures affect t
Microsoft to invest $5.5 billion in Singapore's AI industry through 2029, strengthening position in Asia
Microsoft will invest $5.5 billion in Singapore's AI industry through 2029, doubling down on the country as a key hub for AI development and
Micro1 Collects Home Videos Worldwide to Train Humanoid Robots
To train humanoid robots, startups like Micro1 hire workers in dozens of countries to film household chores on their smartphones — as data g
Kling AI, Runway, and Vidu Gain Users After OpenAI Closes Sora Video Service
After the announcement of Sora's shutdown, users quickly migrated to Kling AI, Runway, and Vidu: Kling's weekly active audience grew by 4%,
How AI Transformed Diary Research: Three Compromises We Left Behind
A UX research team explained how AI made it possible to abandon compromises on sample size, duration, and analysis depth—without sacrificing
Micro1 hires people globally to train humanoids as AI market demands new benchmarks
Thousands of workers are already filming household chores to train humanoids, while researchers warn that current AI benchmarks poorly refle
Marcin Moskala Audited GeminiAI: What the Code Review Revealed About Coroutines and Android Architecture
The author of the open-source GeminiAI client shared how his project passed Marcin Moskala's audit and why coroutine errors became the key t
Coface: artificial intelligence threatens high wages and tax revenues in Britain
Coface rated the United Kingdom as the most vulnerable among 12 developed economies to AI task automation: high-paying roles and their assoc
Anthropic: Claude Code source leak revealed complex agentic architecture
A leaked sourcemap from Claude Code showed that Anthropic's product has long evolved from a 'chat CLI' into a platform with sub-agents, memo