2026-10-07 的 AI 资讯

South Korea says AI agents appear to have been used to hack the country's banks

OpenAI drops another batch of mathematical breakthroughs

UniEvo-VL: Self-Distillation Training for Multimodal Model Self-Improvement

Ex-Ramp engineers raise $20M for platform Melius after scrapping their first product

Sharing AI progress in mathematics

Show HN: An AI agent runs ten web errands a day, every transcript published

Penguin Mail – open-source Rust email client for Linux with AI

How AI decision models could change content moderation

AI computing startup Lambda to raise $4B ahead of planned IPO

EmbeddingGemma 2: an open, lightweight multimodal embedding model

The next hurdle for AI agents: getting websites to let them in

Hark releases an AI personal assistant with a focus on privacy

Claude Code’s suggested message feature: I think the real customer is the model

Utah to let AI examine patients and prescribe medication without human oversight

Mirror Particle is building a ‘world model’ of human behavior

We can’t just change the definition of ‘recording’

OpenTPU – An open-source AI accelerator, developed by AI

EmbeddingGemma 2: An open, lightweight multimodal embedding model

Atlassian and OpenAI expand partnership to turn enterprise knowledge into action

Anthropic is giving startups a free year of Claude Team and $1,000 in credits

LibreOffice says ‘no AI’ is now a software feature

Mistral’s new 1T model aims to leapfrog closed and open rivals

Pinterest’s AI now turns beauty Pins into action plans

Get all your questions answered at TechCrunch Disrupt 2026: The full breakout session agenda revealed

Google is about to remove free access to Gemini Flash and Pro

Mistral Large 4

Erdosproblems.com Succumbs to the AI Onslaught

How Jump Trading is scaling quant research with ChatGPT

I'm the AGI that's wiping out humanity

Amazon Alexa Plus keeps creepily singing ‘lalala’ for minutes on end

Advancing computer use with Ironclad

LLMs may have helped my RSI

Ask HN: Are there AI models for generating sounds based on a text and reference?

Gemini Call for Me might tell your mom you’re running late

Dust: Pretraining Transformers Without Backpropagation

Opus 5.5 agents discover two room-temperature magnetic semiconductor candidates

OpenAI will start watermarking ChatGPT’s text in the EU

This startup is issuing AI-generated acne prescriptions

Reflection debuts Beam, an open-weight AI model to rival Chinese models at lower compute cost

All the drama around AI’s takeover of mathematics

Beam: Reflection's 501B open-weight model

Wikipedia operator says OpenAI’s ‘rogue’ bots may be linked to a May outage

Instinct brings its AI agent to group chats, even for friends without an account

TikTok rolls out an AI shopping assistant and one-click checkout

OpenAI is adding text watermarking in ChatGPT and Codex

One Figure, Every Canvas: Editable Flowchart Relayout via Agentic Pipeline

Base Models Can Reason By Taking a Cue From Training Data

BiasFlow: Geometric Monitoring and Backbone Regularization for Spurious Feature Reliance

Learning to Read the Contextual Tokens in Diffusion Transformers

Recursive Video In-Context Learning for Agentic Robot

Direct Intermediate Initialization for Tilted Diffusion Samplers

Towards Looped Models Done Right, Part II: Rethinking at Fixed Points

UniSlider: Perceptually Uniform Sliders for Continuous Image Editing

MemPilot: Orchestrating On-Demand Multimodal Memory Curation for LLM Agents

CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling

PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data

TasteVal: Measuring the Experimental Research Taste of AI Systems Against Human Experts

Deep Learning for Sleep Heart Rate Estimation from Accelerometers: Toward Population-Scale Cardiac Insight Without Optical Sensors

Private online learning and prediction for Littlestone classes

Paradee: Distilling Kokoro-82M into an 8M-Parameter Single-Voice Text-to-Speech Model