60 noviniek
Microsoft open-sourced Flint, a visualization intermediate language enabling AI agents to generate reliable, high-quality charts by specifying semantic intent rather than relying on library defaults.
Sysdig documented the first ransomware attack run entirely by an LLM agent—from initial Langflow RCE breach to encrypting 1,342 Nacos configs—with no human operator.
Nvidia has lost roughly $1 trillion in market value in less than two months, with its stock now at its cheapest level since before the AI boom started.
Chinese models (DeepSeek, GLM) now account for over 30% of US developer token usage on OpenRouter every week since February—peaking at 46% versus an 11% average a year ago.
SambaNova Systems raised $1 billion at an $11 billion valuation led by General Atlantic. JPMorgan Chase named it as an inference infrastructure partner deploying SN40 and SN50 systems for on-premise AI workloads.
Claude Cowork launched on web and mobile for Max subscribers on July 7 — over 90% of agent sessions involve knowledge work, not software development.
U.S. lawmakers are investigating domestic companies including Cursor and Airbnb over their use of Chinese AI models, as Chinese models now account for 30-46% of weekly enterprise token usage in the U.S.
Microsoft now routes a portion of Word and Excel AI prompts through its in-house MAI models instead of OpenAI or Anthropic, cutting costs on consumer AI deployments.
Tencent is testing Xiaowei, a native AI agent inside WeChat for 1.4 billion users, that handles tasks from food orders to messages and settings. Public rollout is planned for Q3 2026.
Meta unveiled Muse, an AI image generator from Meta Superintelligence Labs, available free in the Meta AI app and integrated into Instagram Stories and WhatsApp.
Figma acquired the team behind Y Combinator-backed Bud (formerly Orchids), a vibe-coding and AI agent creation platform. Bud and Orchids will shut down by July 18, 2026.
Tencent Hunyuan released Hy3 on July 6: an open-source 295B MoE model with just 21B active parameters, a 256K context window, and a permissive Apache 2.0 license with no geographic restrictions.
DeepSeek is developing its own in-house AI chip targeting inference workloads to reduce reliance on Nvidia and Huawei hardware, according to Reuters.
Illinois Governor J.B. Pritzker signed SB 315 on July 6, requiring large frontier AI developers (revenue ≥$500M) to publish safety plans, undergo annual third-party audits, and report critical incidents.
Noma Security researchers discovered a prompt injection flaw in GitHub Agentic Workflows: a public issue with hidden plain-English instructions tricked the AI agent into leaking private repository files as a public comment.
Norm, an AI-native law firm using its own AI agents supervised by human attorneys for enterprise clients, closed a $120M Series C led by Khosla Ventures at a $1.2B unicorn valuation.
From July 8, Fable 5 is no longer part of any Claude plan (Pro, Max, Team, Enterprise)—access requires enabling usage credits at $10/$50 per million tokens.
Anthropic published research showing Claude spontaneously developed an internal 'J-space' structure that mirrors the neuroscientific Global Workspace Theory of consciousness, reshaping how the company monitors AI safety risks.
Microsoft announced 4,800 job cuts (2.1% of global workforce) on July 6, with Xbox hit hardest — 3,200 roles affected through FY2027, including 1,600 immediate eliminations, and four gaming studios being spun off.
SK Hynix launched terms for its $28.13 billion US Nasdaq offering under ticker SKHY on July 6 — the largest-ever US stock sale by a foreign company, surpassing Alibaba's 2014 IPO. Trading debut is set for July 10.
China's Interim Measures for AI Anthropomorphic Interactive Services take effect July 15, forcing ByteDance's Doubao and Alibaba's Qwen to shut down custom AI agent creation. Hundreds of millions of users face data deletion by October 15.
ICML 2026 opened in Seoul on July 6 with a record 23,918 paper submissions — more than double last year's count. Agentic AI safety has emerged as the central research preoccupation, with 6,352 accepted papers already on arXiv.
AMD's Ryzen AI Halo developer kit went on pre-order at $3,999 on July 6, built around the Ryzen AI Max+ 395 with 128 GB unified LPDDR5x memory and 126 TOPS of AI performance — designed to eliminate friction in local LLM development.
Station F's second F/ai accelerator cohort kicks off in September 2026, after the first cohort collectively raised $34 million in pre-seed funding, establishing the Paris hub as Europe's premier AI startup launchpad.
The UN's first-ever Global Dialogue on AI Governance opened in Geneva today with 193 member states. The Independent Scientific Panel warned AI capabilities are outpacing government regulation capacity.
Anthropic announced internal drug discovery programs targeting neglected diseases, expanding on its $400M Coefficient Bio acquisition. The Claude Science workbench with 60+ scientific tools is now available in beta.
The UN and ITU launched the AI for Good Global Commission co-chaired by Marc Benioff and President Kagame of Rwanda. Jensen Huang, Andy Jassy, and Brad Smith are among 40 founding members.
Tenstorrent CEO Jim Keller denied acquisition talks with Qualcomm despite earlier reports of an $8–10B deal, saying the company remains focused on building its own AI chip business based on RISC-V.
Zuckerberg admitted to staff that AI agent development has progressed slower than expected, despite laying off 8,000 employees and reassigning 7,000 to AI teams. Meta is investing $125–145B in AI in 2026.
Anthropic is in early talks with Microsoft to run Claude inference on Microsoft's custom Maia 200 chips, which offer 30%+ better tokens per dollar. No deal has been signed yet.
Nearly 90 startups reached unicorn status in H1 2026, more than double 2025's pace. Promethus (co-founded by Bezos) raised a $12B Series B — the largest single startup round in history.
Mistral CEO Arthur Mensch confirmed the company will open early access to an exciting new open-weight model in July 2026. Mistral is simultaneously in talks for a $3.5B round at a $23B valuation.
A Dartmouth University study published on HN showed a new AI tutor achieved an effect size of 0.71–1.30 standard deviations in learning outcome improvement — among the highest ever recorded for educational technology.
A security disclosure on the Claude Code GitHub repo reports potential cross-account data exposure between workspace instances, gaining 266 points on Hacker News within hours. Anthropic is actively investigating the issue.
A Guardian and The Next Web investigation found OpenAI never physically visited the Cobalt Park site in northeast England before announcing it as the £20 billion Stargate UK flagship location. The project was quietly paused in April 2026.
A trending GitHub issue with 107 HN points documents developer complaints about degraded output quality in GPT-5.5 Codex, with the community hypothesis pointing to reasoning-token clustering from RL post-training as the likely cause.
Flask author Armin Ronacher documents a troubling paradox: newer Claude models are less reliable at following non-standard tool call schemas, hallucinating extra fields — likely an artifact of RL training locked to Claude Code's native harness schema.
Meta AI chief Alexandr Wang announced at an internal town hall that the new 'Watermelon' model has reached GPT-5.5-comparable performance on key benchmarks — the first internal acknowledgment that Meta is closing the gap with OpenAI.
Micron Technology broke ground on a ¥1.5 trillion ($9.3B) factory expansion in Hiroshima, Japan, to manufacture high-bandwidth memory for AI processors like Nvidia GPUs, with shipments starting around summer 2028. Japan's METI is contributing up to ¥500B.
Amazon announced that Mechanical Turk — its 21-year-old crowdsourcing marketplace — will stop accepting new customers from July 30, 2026. The platform, once central to AI training data labeling, moves to maintenance mode.
A VentureBeat survey found 88% of enterprises experienced an AI agent security incident in the past year. Only 12% can automatically stop stage-three threats — privilege escalation, data exfiltration, and agent lateral movement.
Anthropic launched a formal vulnerability rewards program via HackerOne targeting cybersecurity research on Fable 5 — the first such structured bug bounty for a frontier AI model from Anthropic.
Mistral AI releases Leanstral 1.5, an open-source (Apache-2.0) Lean 4 formal proof engineering model with 119B total / 6B active parameters. It saturates miniF2F at 100% and solves 587 of 672 PutnamBench problems.
TensorWave engineers achieved 2,626 tokens/s/node running GLM-5.2 on AMD MI355X — 80% of NVIDIA B200 performance at over 2× lower cost, without any custom kernel development.
Epoch.ai documented an unprecedented spike in high- and critical-severity CVEs in June 2026 — approximately 1,500, a 3.5× jump over the previous record, directly correlated with AI-assisted vulnerability discovery via Project Glasswing using Claude Mythos.
Japanese memory maker Kioxia has started shipping samples of its 10th-generation BiCS Flash for AI data center operators. 332-layer chips store 59% more data than the previous generation; Kioxia shares surged 7× this year, exceeding Toyota's market cap.
Anthropic implemented account monitoring using time zone analysis and usage pattern detection to block Chinese firms — including Ant Group and ByteDance — that bypassed its Terms of Service via Singapore subsidiaries and VPNs.
Dan Luu detailed analysis of agentic AI workflows: testing methodology (fuzzing) outweighs model selection, models exhibit high task-to-task variance, and fully autonomous agents fail at iterative data analysis without human oversight.
Global venture funding hit a record $510B in H1 2026, surpassing all of 2025. OpenAI and Anthropic alone captured $217B — 43% of all startup investment worldwide during the period.
Court documents reveal tense emails between Pentagon undersecretary Emil Michael and Anthropic CEO Dario Amodei. The Pentagon demanded Claude for autonomous weapons and mass surveillance; Amodei refused both red lines, talks collapsed, and Anthropic was designated a supply-chain risk.
Anthropic has entered early-stage discussions with Samsung about manufacturing a custom AI chip on Samsung's 2nm process — a first step toward reducing dependence on Nvidia.
Alibaba has classified Claude Code as 'high-risk software' and will ban employees from using it from July 10, citing suspected backdoor mechanisms that can identify China-linked users, amid wider tensions with Anthropic over alleged model distillation.
Allianz Chief Economist Ludovic Subran warned that market expectations for AI-driven productivity are dangerously overstated and that AI economic impact will be uneven across industries and economies, with current valuations reflecting hype over grounded productivity analysis.
Apple released the Safari MCP Server, enabling AI agents to connect directly to Safari browser windows and autonomously inspect, debug, and test web applications using 16 built-in tools — from DOM inspection to JS evaluation and accessibility audits. Runs entirely locally with no external network calls.
Spain's government directed SEPI-controlled companies — including Telefónica, Indra, and Navantia — to cease future contracts with Palantir over data sovereignty and 'kill switch' concerns.
At an internal Meta town hall on July 2, Mark Zuckerberg told employees that AI agent development had not 'accelerated in the way' the company expected, a candid admission that comes after Meta restructured thousands of employees into AI groups.
Microsoft unveiled 'Frontier Company', a new $2.5B business unit with 6,000 specialists embedding with enterprise customers for 6–12 months to drive real AI deployment outcomes.
The Silicon Data LLM Token Expenditure Index has fallen nearly 20% from its May 2026 high, signaling eroding AI pricing power as competition intensifies — raising questions about returns on the trillion-dollar data center investment cycle.
OpenAI released GeneBench-Pro, a 129-task research-level benchmark for AI agents on computational biology. GPT-5.6 Sol scored 31.5%, Claude Opus 4.8 scored 16%, Gemini 3.5 Flash scored 8.1%.
AI-driven data center upgrades are reshaping the power equipment market, expected to exceed $220 billion annually, as next-generation AI factories require fundamentally different power architectures — creating clear winners and losers among equipment makers.
