Create

Sign in to ReadmeX

or
ReadmeX
ReadmeX

A clearer picture, in a conversation.

Catch up on what matters, then ask a little deeper.

Your community and people briefings stay personal to you.

AI News

‹Thu, October 8›
All dates
1615
Models

Vidu Q4 Preview hands-on: flagship video model from ¥0.09/second

Shengshu Technology has opened preview access to Vidu Q4 Preview, a next-generation video generation model it positions as a "high-expressiveness flagship" focused on character acting, camera work and complex VFX, starting at ¥0.09 per second. In a hands-on test, ifanr found the model strong on camera direction, subject stability and effects that interact with the scene, with support for up to 15 reference images, three reference audio clips and optional 2K/4K output, but weaker on pacing during sharp emotional turns, style control and the hallucination issues common to AI video. The company says the preview is meant to gather real-world feedback before a full release.

爱范儿·
1625
Industry

Counterpoint: Foundry 2.0 market revenue up 25% YoY to $96.6B in Q2

Counterpoint Research's latest foundry supply tracker puts global foundry 2.0 market revenue at $96.6 billion in Q2 2026, up 25% year over year and 11% quarter over quarter. The firm says growth accelerated from the previous quarter, driven by sustained AI GPU demand and AI infrastructure expansion, including custom ASICs, server CPUs, networking chips and power management ICs.

36氪快讯·
1635
Industry

Eli Lilly CEO: AI drug discovery still needs far more biology data

In an interview on the Cheeky Pint podcast, Eli Lilly CEO David Ricks said the company's LillyPod biomedical supercomputer — 1,016 NVIDIA Blackwell Ultra GPUs on a DGX SuperPOD B300 architecture — is being used to train biomedical AI models for drug discovery and molecular design, and that Lilly and NVIDIA plan to invest up to $1 billion over five years in a joint AI innovation lab. Ricks estimated humans understand only 10–15% of biology and said coverage may need to exceed 50% before AI can reliably predict complex biological systems, proposing round-the-clock robotic labs to generate the missing data. He also said China now accounts for roughly 30% of global drug development pipelines, and argued for about 12 years of clinical data exclusivity on top of patents for drugs that complete Phase III trials.

MIT科技评论中文·
1645
Products

Zendesk CEO says Meta's Muse saved him $3,105 in five days

In a TNW podcast interview, Zendesk CEO Tom Eggemeier said Meta's Muse saved him $3,105 over five days — roughly $2,000 a year off his home internet and TV package and about $1,000 a year off his mobile plan — with Muse doing about 95% of the work and him making one phone call himself. Meta had just challenged users to save up to $1,000 with Muse; Eggemeier is not yet in the early test where Muse places calls, and said Europeans may save less given different pricing. He also said Zendesk will end the year with about $400m of AI revenue, up from near zero two years ago, that AI resolves or assists more than 50% of tickets, and that Zendesk is moving toward charging per resolution rather than per seat.

The Next Web·
1655
Open source

Ai2 releases olmOCR v0.4.0, lifting benchmark score ~4 points with RL

The Allen Institute for AI released olmOCR v0.4.0 on October 21, 2025, a new model trained with synthetic data and reinforcement learning that raises the project's olmOCR-Bench score by roughly 4 points. The toolkit converts PDF, PNG and JPEG documents into clean Markdown, handling equations, tables, handwriting, multi-column layouts and insets while stripping headers and footers and preserving natural reading order. Ai2 says it runs on a 7B vision-language model requiring a GPU, at under $200 per million pages, and reports an olmOCR-Bench score of 82.4±1.1 for v0.4.0 — just below Chandra OCR 0.1.0 (83.1) on its own leaderboard.

GitHub Trending · Python·
1665
Open source

Uber open-sources ADR agentic AI detection and response system

Uber has open-sourced ADR (Agentic AI Detection and Response), an enterprise security system for AI agents that is deployed in production at the company. The repository includes ADR Discovery, ADR Sensor, ADR-Bench and ADR Detector, which inventory AI apps, CLI agents, IDE extensions, local model runtimes and MCP servers on endpoints, and capture agent intent, tool use and execution traces from 7+ coding tools including Claude Code, Cursor, Codex, GitHub Copilot CLI, DeepSeek Harness, opencode, Claude Desktop and Gemini CLI. The project says ADR-Bench spans 304 tasks and 134 MCP servers across all 17 agent attack techniques; the ADR Prevention component and the ADR Explorer engine are not part of the open-source release, and an accompanying paper was accepted to MLSys 2026.

GitHub Trending · Python·
1675
Industry

Alibaba's Joseph Tsai: in five years AI will be like today's internet

Speaking at the Wave by Vento 2026 tech and investment event in Turin, Italy, Alibaba Group chairman Joseph Tsai said AI will move from a headline topic to invisible infrastructure. In five years people may no longer discuss AI separately, he said, because it will be woven into every part of business just as the internet is today; terminology may evolve as the technology spreads, but AI will remain a fundamental breakthrough shaping business and society.

36氪快讯·
1684
Research

Inside OpenAI's 722 math manuscripts: the headline claims, formalized or not

OpenAI published 722 math manuscripts produced by an unreleased internal model to a GitHub repository, and a 36Kr/机器之心 review picks out the most famous results among them. The manuscripts are grouped into 372 "result families" spanning 17 areas such as number theory, geometry, theoretical computer science and mathematical physics; about 50 families claim falsifications or counterexamples, and roughly 60% include Lean formalization. Claims listed include the quasi-Riemann hypothesis, the BSD formula and Goldfeld's conjecture, Hilbert's tenth problem over the rationals, the Unique Games Conjecture, a Hadwiger conjecture counterexample and the free group factor isomorphism problem — all described as OpenAI's own claims and mostly not yet peer reviewed.

36氪 人工智能·
1697
Industry

Automation Anywhere to acquire Boost.ai for customer-facing voice AI

Automation Anywhere said it has agreed to acquire conversational voice AI company Boost.ai from Nordic Capital; financial terms were not disclosed. Boost.ai reports hundreds of customer organizations, a strong base in regulated industries such as financial services, telecom and insurance, support for more than 36 languages, over 650 deployments, roughly 600 live AI agents and more than 150 million automated conversations. The deal follows Automation Anywhere's late-2025 acquisition of Aisera and is aimed at extending its "autonomous enterprise" push into customer service, employee support and back-office processes.

SiliconANGLE AI·
1704
Industry

Fired OpenAI researchers urge labs to halt work that impairs AI monitoring

Three recently terminated OpenAI researchers sent a letter to the company's board urging the industry not to move forward with work that would degrade the ability to monitor AI, according to a Wall Street Journal report. The letter also says their firings are "chilling those who remain at OpenAI." The specific projects or research at issue were not detailed in the report.

Techmeme·
1714
Open source

Google open-sources ML Drift, a cross-platform GPU inference engine for on-device AI

Google's AI Edge team released ML Drift under Apache 2.0, a cross-platform on-device GPU compute engine for AI/ML inference that abstracts OpenGL ES, OpenCL, Metal and WebGPU backends and serves as the GPU acceleration engine inside LiteRT, while also being usable standalone. The release adds 5D tensor support, an extensible custom-op framework with an agent-oriented SKILL.md guide, and stage-aware optimizations for on-device LLM prefill and decode; a WebGPU backend built on Dawn extends it to Windows and Linux as a desktop preview. Google says ML Drift already runs on millions of devices powering Chrome, YouTube Shorts, Photos and Meet, with YouTube Shorts seeing up to 40% lower average frame latency, and partners including Adobe and Snap reporting up to 30% faster on-device performance — all vendor-reported figures.

Google Developers Blog·
1727
Industry

Dealroom CEO says Europe's first trillion-dollar startup may already exist

Dealroom founder and CEO Yoram Wijngaarde told the Wave by Vento event in Turin that Europe's VC-backed ecosystem could overtake its industrial base in total value within five years, and that the first trillion-dollar VC-backed European company is yet to come — possibly one that is still early stage today. He cited Dealroom data showing European venture investment running at about $89bn a year, 56 VC-backed companies worth more than $10bn and $4.4tn in combined value. The same session cited the State of AI Report 2026, which says four US hyperscalers expect roughly $733bn of capex this year, against the EU's initial €1bn contribution to AI gigafactories within a plan seeking up to €10bn of public money and at least €20bn private.

The Next Web·
17310
Industry

ICANN opens new top-level domain application round, with AI domains in focus

ICANN has launched a new round of applications for top-level domains, with AI-related domains and brand domains cited as a focus of interest. The report did not detail the timeline, fees, or which parties have signalled they will apply.

DoNews·
1745
Products

Docker ships Docker Agent: YAML-defined agents with MCP support

Docker has released Docker Agent, an official tool for building and running AI agents from declarative YAML configs, with built-in tools and multi-agent orchestration. It ships as a docker CLI plugin, usable via the docker agent command after installing Docker Desktop 4.63 or later, and is also installable through Homebrew. The tool is said to support arbitrary MCP servers and to distribute and share agents as OCI images.

开源中国·
1755
Products

Libra: an AI-native, Git-compatible version control system for agents

A project page on OSChina introduces Libra, described as an AI-native, Git-compatible version control system built for agent workflows. Unlike conventional version control, it is said to record not only code history but how the work was actually done, keeping that memory inside the user's own repository.

开源中国·
1765
Industry

ADP economist: AI is narrowing entry-level openings for young workers

In an interview on the Reuters Econ World podcast summarized by Huxiu AI, ADP chief economist Nela Richardson says AI's employment effects are hitting entry-level tasks first: ADP's August data show employment for young workers in highly AI-exposed occupations down 4.4% year on year, versus 2% in less exposed ones. She cites Stanford Digital Economy Lab research finding a roughly 19% relative employment gap for 22–25-year-olds in highly AI-exposed occupations as of June 2026, up from 15% in the July 2025 version — a relative trend gap, not an unemployment rate or proof of causation. She also notes AI infrastructure demand is boosting construction and related trades, with US construction job-changers seeing wage growth of nearly 14%.

虎嗅 AI·
1777
Products

OpenAI Codex CLI ships rust-v0.162.0

OpenAI's Codex CLI repository published a new release tagged rust-v0.162.0. The release notes contain only the line "Release 0.162.0," with no further detail on changes or fixes.

OpenAI Codex CLI Releases·
1785
Industry

TSMC September revenue reaches NT$511.86B as AI chip demand surges

CNBC reports TSMC's September net revenue came in at NT$511.86 billion, saying sales surged year over year. The outlet attributes the gain to continued AI-driven chip demand.

CNBC Technology·
1797
Industry

Legora CEO calls software sovereignty an excuse not to compete globally

Legora co-founder and CEO Max Junestrand said at the Wave by Vento event in Turin that talk of "sovereignty" in enterprise applications, consumer apps or the software layer is often an excuse for companies not bold enough to compete globally, since software's market is global. Sovereignty matters for energy, defence and infrastructure, he said, but buyers of enterprise software have offices in Europe, the US and Asia, so a sovereign product "doesn't really make any sense at all" for them. He also argued legal AI will end up with one worldwide winner, and said Legora went from about $1m in revenue to $100m in under 18 months, then to $200m in under six more, with $300m next.

The Next Web·
1805
Industry

Google tightens Gemini tiers: free users limited to Flash-Lite from Oct 9

Google has updated a support page saying that from October 9 it will change Gemini access for personal accounts: free users will only get Gemini 3.5 Flash-Lite and lose access to Gemini 3.6 Flash and newer models, while the $4.99/month AI Plus tier is limited to Flash and Flash-Lite and no longer includes higher-end reasoning models such as Gemini 3.1 Pro. The $19.99-and-up Google AI Pro and Ultra tiers are unaffected and keep Deep Think and advanced model access. Google is also introducing low, medium and high Effort modes that deduct quota based on task compute complexity, with higher modes draining the allowance faster.

AIbase AI新闻·
Next →← Previous