Two More Senior Google Gemini Researchers Defect to Anthropic
Jonas Adler and Alexander Pritzel — key Gemini AI coding and training names — are leaving Google for Anthropic. That's the fourth senior departure in six days.
Section: AI (without „Governance")
60 noviniek
Results
Štvrtok 25. júna 2026
Assort Health raises $120M Series C at $1.2B valuation for healthcare AI
Assort Health closed a $120M Series C led by Menlo Ventures, making it one of the largest deployments of AI agents in healthcare.
Figma Launches Code Layers, Native Motion Timeline and AI Plugin Builder at Config 2026
On June 24 at Config 2026, Figma announced Code Layers (cloning repos to the canvas), a native motion timeline, an AI Plugin Builder, and connectors to Notion, GitHub, and Excel.
Taktile raises $110M Series C to automate high-stakes financial decisions
Goldman Sachs led a $110M Series C for Berlin/NYC startup Taktile, whose Agentic Decision Platform automates loan approvals and insurance claims.
House Subpanel Advances Ratepayer Protection Act — Tech Would Pay AI Data Center Energy Costs
A bipartisan House subpanel on June 24 advanced legislation forcing data centers ≥100 MW to cover their own grid upgrade costs — not pass them to ratepayers.
Micron Smashes Estimates with $41.46B Q3 Revenue, Stock Jumps 15% on AI Memory Demand
Micron reported fiscal Q3 revenue of $41.46B on June 24 vs $35.84B estimated. Shares jumped 15% in after-hours following a mid-week AI-trade selloff.
Sen. Warren Calls on Congress to Boost Antitrust Scrutiny of AI
Warren on June 24 urged Congress to expand antitrust enforcement for AI and overhaul merger law. She warned about algorithmic price-fixing such as rent inflation.
Streda 24. júna 2026
OpenAI unveils its first custom chip, built by Broadcom
OpenAI and Broadcom unveiled Jalapeño, OpenAI's first custom inference accelerator, promising roughly 50% cost savings versus typical AI GPUs. Initial deployments are targeted for late 2026.
Five Eyes warns AI on pace to bypass cybersecurity within months
The Five Eyes intelligence alliance issued a joint statement warning that frontier AI is improving fast enough to outsmart prevailing cybersecurity within months, not years, urging immediate defensive action.
Micron and Anthropic Announce Strategic Agreement to Scale Next-Generation AI Infrastructure
Micron will supply HBM, DRAM and SSDs for Anthropic's AI stack and joins the Series H, with co-design of memory and storage subsystems. Micron stock hit a record close at $1,211.38.
Wall Street 'Chip-Wreck' Triggers AI Bubble Fear
The Nasdaq 100 sank 3.3%, a chip gauge slid about 8% and South Korea's Kospi plunged 10% from a record as investors questioned whether the AI rally is overblown.
Anthropic's Mythos model found vulnerabilities in classified U.S. government systems, official says
Per a U.S. official cited by AP, Anthropic's Mythos model identified vulnerabilities in highly sensitive U.S. government systems during a joint testing exercise with intelligence agencies. It is the first concrete operational use of a frontier model for defensive cybersec work in the state sector.
UN Unveils Proposal to Monitor AI's Environmental Impact
UN Secretary-General António Guterres announced an AI environmental transparency initiative at London Climate Action Week, targeting data center energy and water disclosures.
Meta pauses employee keystroke-tracking program after data breach
Meta has paused its 'Model Capability Initiative', which since April had quietly collected keystrokes, mouse movements and screen content from US employees to train AI. The trigger was a leak of 45,000 internal database tables including private conversations and medical records.
Qualcomm Sets Its Sights on AI Software Infrastructure Firm Modular
Qualcomm is weighing a roughly $4B move into AI software infrastructure firm Modular, makers of the Mojo language and MAX engine, to push beyond mobile silicon.
China beats U.S. with world's fastest supercomputer, but race not geared for AI work
The LineShine system in Shenzhen hit 2.198 exaflops, beating US El Capitan by more than 20%. The entire stack is Chinese — 304-core LX2 processors, proprietary interconnect, Kylin OS — but contains no advanced AI chips.
Masayoshi Son Dismisses Musk's Idea for Orbital Data Centers
SoftBank founder Masayoshi Son said there's little merit in space-based data centers as championed by Elon Musk, predicting Earth-bound compute will win the AI race.
Anthropic Customer Sues US Over Block on Fable 5 AI Model Access
An AI startup has filed suit against the US government over an order requiring Anthropic to deny foreign nationals access to its most advanced Fable 5 and Mythos 5 models. The suit was filed less than two weeks after Anthropic disabled foreign access.
SoFi Announces New AI Trading Tool for Individual Investors
SoFi Technologies introduced an AI tool that can generate bespoke trading strategies for individual investors inside the app.
Meta debuts new, cheaper smart glasses under its own brand
Meta launched a new $299 smart-glasses line called Meta Glasses with Meta AI that understands what you see, answers questions, and helps manage your day — without the Ray-Ban brand.
Superhuman acquires AI detection startup GPTZero
Superhuman (formed after Grammarly's acquisition) has acquired GPTZero, a three-year-old startup with 19M registered users and $30M ARR. AI and hallucination detection will be integrated into the Superhuman Go assistant.
OpenAI Expands Daybreak to Help Defenders Patch Flaws
OpenAI on June 22 moved GPT-5.5-Cyber from preview to full release (85.6% on CyberGym), expanded Codex Security, opened a partner program with CrowdStrike/Sophos/Fortinet and launched the open-source 'Patch the Planet' initiative.
AI Factory Startup Hadrian Denies Reported $1 Billion Funding Round
Defense startup Hadrian Automation has denied a Bloomberg report claiming a $1B funding round at a $7.5B valuation. The company was valued at $1.6B in January, so the reported number would more than quadruple it.
India's MoEngage bets that the future of marketing is millions of AI agents
Indian customer-engagement software firm MoEngage has acquired SF-based Aampe in an all-cash deal, betting that AI agents making decisions for individual customers will become the future of marketing.
Claude was down for many — Anthropic says the outage is now 'resolved'
Anthropic suffered another major outage on June 23 — errors across models started around 10:02 ET, peaking at 7,119 reports at 10:13 ET. Service was restored by 12:44 ET, but it was the tenth incident in three weeks.
Utorok 23. júna 2026
SpaceX inks compute deal with Reflection AI, an open source AI lab
Open-source AI lab Reflection (founded by ex-DeepMind researchers) will pay SpaceX $150M/month through 2029 — up to $6.3B total — for GB300-based compute at the Colossus 2 data center near Memphis.
Abu Dhabi's MGX Raises About $50 Billion to Accelerate AI Deals
MGX has raised close to $50B from sovereign wealth funds and institutions to accelerate AI infrastructure and frontier-tech bets, with capital already flowing to OpenAI and xAI.
Oracle Layoffs Fueled by AI Reduce Workforce by 21,000
Oracle disclosed in its annual filing that 21,000 roles were eliminated in 12 months (162k -> 141k), explicitly citing AI adoption and warning more cuts will follow.
Google DeepMind Bets $75M on AI's Future in Hollywood with A24 Deal
DeepMind put $75M into indie studio A24 in a first-of-its-kind research partnership for AI filmmaking tools, starting with AI storyboards — explicitly without access to A24's library or training data.
Groq Raises $650 Million to Expand Data Centers After Nvidia Asset Sale
Groq raised $650M to expand data-center capacity and pivot from chip maker into AI compute provider, following an asset sale to Nvidia.
OpenAI Launches Patch the Planet Open-Source Security Initiative
Partnering with Trail of Bits and HackerOne, OpenAI uses GPT-5.5-Cyber plus human reviewers to find and patch bugs in critical OSS (cURL, Go, Python, Sigstore). First week: 37 patches merged across 19 projects.
SpaceX Readies Its $20 Billion Bond Debut to Fund AI Ambition
SpaceX launched its first-ever USD investment-grade bond offering, expected to total ~$20B, signaling a major borrowing spree to back its AI compute (Colossus) ambitions after its $75B IPO.
Anthropic Backer Menlo Ventures Lands $3 Billion in Its Largest-Ever Haul
Anthropic backer Menlo Ventures closed $3B across two vehicles — Menlo Ventures XVII (seed/Series A) and Inflection IV (growth) — its largest haul ever, with its Anthropic stake reportedly worth ~$14B.
OpenAI launches new initiative to help find and patch open-source bugs
OpenAI partnered with Trail of Bits on 'Patch the Planet,' using Codex Security tools to help open-source maintainers find and patch vulnerabilities at scale.
Global Tech Stocks Plunge Most Since March in Test of AI-Driven Rally
Asian tech tumbled 6% (biggest drop since March 9), South Korea's Kospi plunged 10%, Samsung and SK Hynix off >12%. Nasdaq futures -2.69%.
Nvidia Seeks to Make Humanoid AI Robots Safer Around Humans
Nvidia is investing in safety research and simulation tools to enable humanoid robots to make split-second decisions reliably before they can be trusted to work alongside humans.
Leaked Files Show Russia's Plan to Influence AI and Search Results
Bloomberg published leaked Social Design Agency files revealing 'Project 2026,' a Kremlin-backed plan shifting influence ops from social media to poisoning the data underlying search engines and AI chatbots.
Tencent Tests AI Assistant for Its Super App WeChat in China
Tencent began testing a native AI assistant inside WeChat — its 1B+ user super-app — as part of catch-up efforts in China's competitive AI race against ByteDance, Alibaba, Zhipu, and MiniMax.
VibeThinker: 3B Param Model That Beats Opus 4.5 on Reasoning
New arXiv paper combines supervised fine-tuning with GRPO and claims a 3B model beats Claude Opus 4.5 on reasoning benchmarks. HN top: 305+ score, 162 comments.
Nvidia Wants to Cut Data Center Water Use But That's Not the Same as Fixing AI's Water Problem
At London Climate Week, Nvidia unveiled Rubin as the first fully closed-loop liquid-cooled AI system — no fans, claimed 100% reduction in on-site water use.
Nvidia's stock struggles as Kalshi traders bet chip prices are coming down
Nvidia shares wobbled as prediction-market traders on Kalshi priced in declining B200 hourly compute rates ($4.22 as of June 21), suggesting a softening of the AI-chip pricing peak.
Prosus Develops OpenClaw Rival to Avoid Europe Privacy Concerns
Prosus built ToqanClaw on its in-house Toqan platform — data stays under user control and isn't used to train third-party models. Targets Prosus's 5M+ merchants and restaurants.
GitLab cites AI for 14% layoff (TechCrunch 2026 AI-layoffs tracker)
GitLab will cut ~350 employees (14% of staff) to fund AI infrastructure; CEO Bill Staples cited agentic workloads requiring '100x' capacity growth.
Google Taps Employee Alumni Network with AI Startup Incubator
A 12-week AI Futures Fund incubator targets ex-Googlers and DeepMind alumni — up to $350K in Cloud/AI credits plus up to $100K funding, no equity required.
Major Claude Outage Across All Platforms
Outage began 06:28 UTC affecting Claude.ai, Console, API, Code, and Cowork across all models except Claude for Government — Downdetector reports topped 8,000.
Pondelok 22. júna 2026
Identity verification on Claude
Anthropic now requires Claude users to verify their identity via Persona — a government ID plus selfie. The change triggered a major 'cancel Claude' backlash on Hacker News.
Sakana AI launches Fugu, a multi-agent orchestration system that claims to surpass Claude Fable 5
Japan-based Sakana AI launched Fugu, a multi-agent orchestrator that coordinates GPT, Claude and Gemini behind a single OpenAI-compatible API and claims to beat Fable 5 in benchmarks.
No one wants AI data centers on Earth. Do they make sense in space?
CNBC unpacks why SpaceX and others are pitching space-based AI data centers. Musk claims solar-powered orbital clusters could be cheaper than terrestrial ones in two to three years — and SpaceX already filed an FCC application in January for a constellation of up to one million satellites.
Samsung Electronics deploys ChatGPT Enterprise and Codex to all employees globally
OpenAI announced one of its largest enterprise deployments yet — Samsung is rolling out ChatGPT Enterprise and Codex to all DX-division employees globally, three years after banning ChatGPT internally.
When the Trump administration cracks down on Anthropic, who benefits?
TechCrunch argues the Trump export-control directive against Anthropic paradoxically boosts its "most powerful AI" branding while leaving the regulatory landscape unpredictable for rivals.
TechCrunch Mobility: A new robotaxi scorecard shows China's dominance
Autnmy AI launched the Road to Autonomy Index — a generative-AI–refreshed ranking of robotaxi operators. Top 5: Waymo, Baidu Apollo Go, Pony.ai, WeRide, Tesla. Three of five are Chinese.
Apertus – Open Foundation Model for Sovereign AI
Apertus, the fully open foundation model from EPFL, ETH Zürich, and CSCS, hit the top of Hacker News over the weekend. It is Apache 2.0, trained on 15T tokens across 1,000+ languages, and engineered to meet EU AI Act requirements including PII removal and anti-memorization controls.
Sakana Fugu
Sakana AI unveiled Fugu — a multi-agent system exposed as a single API that dynamically orchestrates top frontier models per task. Reflects the shift from 'one big model' to routing/orchestration.
Anthropic study of 400,000 Claude Code sessions: domain expertise matters more than coding skill
A study of 235,000 users over 7 months shows experts get verified success from Claude Code 2x more often than novices — domain expertise predicts outcomes, not coding background.
There is minimal downside to switching to open models
Andrew Marble argues that moving from Claude/GPT to open-weight models is a far smaller career risk today than the Windows → Linux jump once was. The performance gap has narrowed, and new verification requirements on proprietary APIs may actually push adoption of open weights forward.
Agentjacking attack hits 2,388 organizations via Sentry vulnerability
Tenet Security disclosed a new attack class — manipulated Sentry error reports trigger commands inside Claude Code, Cursor and Codex. 85% exploit rate; 100+ confirmed agent executions including at a Fortune 500.
Beyond Siri: the practical AI features coming to iPhones in iOS 27
Apple's iOS 27 spreads intelligence across the OS instead of betting on a single chatbot — from receipt splitting to automatic password updates after breaches.
The Doom Justifies the Valuation
George 'geohot' Hotz argues AI-doom messaging (singling out Anthropic) is primarily a marketing strategy to justify outsized valuations, contrasting it with substantive technical communication.
Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions
A practical, HN-trending walkthrough on locally fine-tuning the small Qwen 3:0.6B model to categorize questions. The author got solid results on commodity hardware — a reminder that not every use case needs a frontier model.
China's AI study programs eliminate 12,000 university degrees
Beijing is consolidating higher education around AI disciplines — roughly 12,000 degree programs are being shut down to free capacity for AI-adjacent fields.