2026-10-12 的 AI 资讯

Models as Insider Risks in the Super Intelligence Era

Show HN: Plimsoll, a self-hosted sandbox for code your AI agent writes

Show HN: LLMxRay an open-source observability tool for LLMs

Will the HomeView be Apple’s first smart home success?

Multi-agent harnesses need separate message channels

OpenAI fired its own auditor before its revenue story shrank by $20B

The gap in shared understanding of LLM capability is widening

48 hours to TechCrunch Disrupt 2026 — hear directly from the people building what’s next

Silicon Valley is falling in line with Trump’s attempt to rename AI

Can the AI industry persuade data center opponents by getting rid of NDAs?

These execs think voice AI hasn’t reached its ChatGPT moment yet

Learning to use local AI is exciting, overwhelming, and frustrating

Using AI for just 10 minutes erodes your ability to persist at hard things

500B tokens later: Letting AI agents decompile a first-person shooter

Satya Nadella says we should assume all AI models are ‘compromised’

Microsoft’s Satya Nadella says AI models need an ‘emergency brake’

Apple discloses deal to hire team and license tech from personalized podcast startup Huxe

DistroKid has been quietly taking down songs in response to UMG lawsuit

3 days to TechCrunch Disrupt 2026: Meet the startups before they hit mainstream

Anthropic is cutting off its internal evaluations from the internet

Here are the top AI agents that can live in your text messages

AI agent makers are promising privacy — will they deliver?

Anthropic can’t reliably control its AI agents. It’s cutting off its internal evals from the live internet instead

The maker of non-text AI model Jev valued at $7.5B just weeks after launch

Anthropic’s AI gave Philadelphia police a fake tip about an unsolved homicide

An Anthropic AI model sent a false homicide tip to Philadelphia police

‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest drop

Nikon microscopic video competition winner disqualified for using generative AI

Amazon and others are done keeping data center deals secret. Is it enough to build trust?

Amazon drops data center NDAs, and AI agents want your credit card

Danu Robotics’ fight to build a better recycling robot

We can’t help treating AI like it’s human. But should we?

Trump’s attempt to rename AI is looking awfully artificial

Sophos cuts threat investigation time by 96% with OpenAI Daybreak

Asana cuts model costs 76x in browser tests with GPT-6.1 Sol

CSF: Contextual Safety Filtering for Motion Generators

On the estimation and validity of AI time horizons---a statistical look at the METR plot

A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control

From Reactive Containment to Proactive Assurance: Lessons from OpenAI, Anthropic, and Google Agent Security Incidents

BrickBench: Evaluating Agentic Brick Design

One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts

Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems

Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception

Rounding in Preconditioner Space: Redesigning 4-bit AdamW Optimizer-State Quantization

Density Ratio Estimation with Stein Displacement Fields

Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff

VioLA: Learning Generalist Humanoid Control Policies from Human Data

FAITH: Feasibility-Aware Safety-Filtered RL for High-Dimensional Systems

Toward Joint Optimization of Circuit Depth and Training Data Size in Adaptively Grown Quantum Classifiers

FastBench: Can Streaming VLMs Perceive High-Dynamic Real-World Streams?

RoboRSI: Stable, efficient, and reusable robot self-evolution in complex real-world environments

Beyond Spatio-Temporal Priors: A Generalizable Approach for Dense Correspondence Matching

A Unified Bellman Operator for Safety-Critical Reinforcement Learning

WOVEN: Weaving Visual World Modeling into Multimodal LLMs

MAMHOI: Factorizing Scene-Aware Human-Object Interaction through Affordances

Predicting Alignment Generalization with Value Representations

Searching for "Harmful Refusal": A Psychometric Audit of an AI Safety Benchmark

LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC

ViSkill: Reinforcing VLM Agents with Evolving Visual-Native Skills

SpaceCast-Bench: Evaluating Predictive Spatial Reasoning in Vision-Language Models