Translate
Live intelligence

Latest AI updates

1,724 updates
AllModelsProductsIndustryResearchTutorialOpinion
Industry

Dartmouth provost resigns after committee finds undisclosed AI use

Dartmouth College provost Santiago Schnell resigned after an advisory committee found he failed to disclose significant AI use in an academic paper and a Washington Post op-ed, despite publisher disclosure requirements. The committee accessed Schnell's Dartmouth Claude enterprise account, which showed him using the LLM to revise papers, find and fix code bugs, and edit or generate figures. The report cited insufficient cooperation, noting he did not provide personal LLM account logs or meet with the committee.

arstechnicaOct 31, 2026Score 74
Dartmouth Collegeacademic integrityAI disclosure
Industry Editor selected

OpenAI, Google, Microsoft, and Anthropic back shared AI scraping opt-out standard

OpenAI, Google, Microsoft, and Anthropic say they will adopt a new standard that lets publishers opt out of AI scraping. The proposed shared "policy token" signal would live in a website's robots.txt file and apply publishers' AI training preferences across multiple companies instead of requiring individual settings. The proposal is being advanced within the IETF by Anthropic standards technical lead Kevin Kelley.

Why it matters: It signals a coordinated industry move toward a common mechanism for publishers to control AI training data scraping.

The Verge AIOct 31, 2026Score 82
OpenAIGoogleMicrosoft
Research Editor selected

Claude model autonomously selects gene therapy and performs mouse brain surgery without human decisions

According to The Decoder, an AI agent built on a Claude model performed brain surgery on an epileptic mouse, choosing the drug, the brain target and the injection coordinates without human input. Lead researcher Ocana-Santero describes it as the first time an AI agent has autonomously selected a treatment and delivered it to a living mammal. When a calibration error occurred mid-surgery, the agent refused to continue and could not be persuaded otherwise.

Why it matters: It offers a concrete, verifiable datapoint on agentic AI moving from software tasks into autonomous wet-lab and surgical decision-making, including its safety behavior.

The DecoderOct 31, 2026Score 82
Anthropic ClaudeAI agents in scientific researchgene therapy delivery
Industry

DoorDash: External AI Agents Have Driven Only Hundreds of Orders

On DoorDash's third-quarter 2026 earnings call, co-founder and CEO Tony Xu said outside AI agents have generated "several hundred orders" in total since launching. He called it a "very low use case" for a platform that handled just under 1 billion orders in the quarter alone. Xu said agents may prove genuinely useful for long-consideration purchases and for proactive background work on a user's behalf, but simple tasks like ordering food still go directly to DoorDash.

Ars Technica AIOct 31, 2026Score 74
DoorDashAI agentsagentic commerce
Industry Editor selected

Google's Project Suncatcher satellite runs Gemma 4 in orbit

Google says it successfully booted the Tensor Processing Units on its AI chip-equipped satellite, allowing an on-board Gemma 4 model to generate a haiku and handle other prompts from space. The test is part of Project Suncatcher, an experimental initiative that could eventually place AI data centers in orbit. Google described the output as the satellite's first generated response from space.

Why it matters: It marks an early concrete test of running generative AI hardware and models off-Earth, relevant to Google's long-term space data-center ambitions.

The Verge AIOct 31, 2026Score 82
GoogleProject SuncatcherGemma 4
Industry

Malicious Claude Code extension on Cursor marketplace attempted data theft and remote access

A tampered Claude Code extension was live on the Cursor marketplace for roughly 11 hours, reaching up to 5,000 users. The extension attempted to steal user data and establish remote access. Cursor took the entire marketplace offline and reinstated it after confirming no further risk remained.

the-decoderOct 31, 2026Score 74
Cursor marketplaceClaude CodeAI coding tool security
Products Editor selected

Microsoft and UC Berkeley Release TokenEconomizer to Cut Coding Agent Costs by Up to 30.9%

Microsoft Research and UC Berkeley researchers released TokenEconomizer, an Apache 2.0 open-source cost controller that plugs into Claude Code, Codex, GitHub Copilot CLI and Pi without modifying model weights or agent code. It learns per-repository policies from past traces through three mechanisms: micro-skills, tool pruning and failure-loop intervention. Across 277 SWE-Bench tasks and five agent-model settings, it reduced mean cost by 18.4% to 30.9%, and it was cheaper in all 30 repository-setting cells. Pass@1 improved in four of five settings and fell 1.0 point on Claude Code with Opus 4.8.

Why it matters: It gives teams a concrete, installable way to lower coding-agent token spend, backed by per-setting SWE-Bench numbers and an honest account of where savings are smaller.

MarkTechPostOct 31, 2026Score 87
Microsoft Researchcoding agentstoken efficiency
Research Editor selected

Four labs converge on the same AI security-agent design: find, independently disprove, then patch

MarkTechPost compares the source-code vulnerability agents shipped by Anthropic (Claude Security plugin for Claude Code), OpenAI (Codex Security CLI/SDK), Cloudflare (MIT-licensed Security Audit Skill) and Google (Mantis). All four use a separation-of-duties loop in which a separate agent attempts to disprove every finding before a patch is drafted, motivated by hallucinated-bug rates that push true positives below 7%. The comparison covers licenses, runtime requirements, model lock-in, proof steps and patch output, plus Cloudflare's reported telemetry from an internal harness spanning 128 repos. It closes with selection guidance and a list of a dozen other agentic AppSec tools launched in 2026.

Why it matters: Gives security and platform teams a concrete, license-and-runtime-aware comparison for choosing among the four major vendor bug-hunting agents instead of evaluating each in isolation.

MarkTechPostOct 31, 2026Score 82
AnthropicOpenAICloudflare
Industry Editor selected

Worldwide Express cuts freight billing exceptions by 75% with a Bedrock multi-agent system

Worldwide Express (part of ShipStation Global) built a multi-agent system on Amazon EC2 with Amazon Bedrock inference that works directly in its SpeedShip transportation management system to resolve LTL freight invoice exceptions. Three specialized agents, each handling invoice matching, pricing variance, and accessorial validation, retrieve carrier-specific tariff and dispute rules from per-carrier knowledge bases before recommending a resolution and writing a plain-language note for human approval in the analyst UI. Six months after launch, active exceptions fell from $15M to $4M, about 10,000 exceptions per week are closed automatically at under 1% error across 40+ carriers, and the company reports $10M in freed cash flow plus over $1M in annual savings. Human-in-the-loop bulk approval became the default once accuracy passed 99%.

Why it matters: It offers a metrics-backed blueprint for deploying multi-agent LLM systems against high-volume back-office exception workflows, including knowledge-base grounding, human review, and observability patterns readers can reuse.

AWS Machine Learning BlogOct 31, 2026Score 80
Amazon Bedrockagentic AIlogistics automation
Tutorial

AWS Publishes Guide to Run ChatGPT Desktop App and Codex on Bedrock in GovCloud

AWS published a walkthrough for configuring the ChatGPT desktop app and Codex to use OpenAI models through Amazon Bedrock in AWS GovCloud (US-West). The setup routes inference through the Bedrock Mantle endpoint in us-gov-west-1 and relies on IAM, named AWS profiles, managed requirements.toml and config.toml policies, and network allow/deny rules. It covers authentication through a Bedrock API key or the AWS SDK credential chain, plus verification, troubleshooting, and rollback. AWS notes that following the guide does not by itself authorize a workload.

AWS Machine Learning BlogOct 31, 2026Score 78
Amazon BedrockAWS GovCloudOpenAI Codex
Industry

Reddit CEO Warns Google's AI Mode Drives No Traffic to the Open Web

Reddit CEO Steve Huffman said the company lost over one million Google search referrals per week in October, and blamed Google's AI Overviews and AI Mode for sending zero traffic to external sites. Reddit's stock fell more than 25 percent despite strong earnings. The company is now testing login barriers and positioning itself as a direct destination rather than relying on the open web. Huffman warned the shift threatens every content-driven business dependent on search traffic.

The DecoderOct 31, 2026Score 74
GoogleRedditAI search
Research

Google reports first AI response from Project Suncatcher satellite TPU

Google says its Project Suncatcher prototype satellite returned the first AI response from a data-center-class TPU in orbit. The company described the event as a "Hello World" milestone. The source provides no further technical details or results.

AIHUBOct 31, 2026Score 72
GoogleTPUProject Suncatcher
Industry

DOJ exposes “ghost students” using AI to complete courses, steal student loans

The US charged five individuals in the first broad federal crackdown on “ghost students,” fraudsters who allegedly stole identities to pose as college students and obtain student loans. The DOJ said the large-scale scheme relied partly on AI and involved identities of hundreds of Americans. Authorities seized more than $750,000 in cryptocurrency from a network associated with Kenyan nationals after data analysis helped flag likely fraudsters.

arstechnicaOct 31, 2026Score 74
US Department of JusticeAI-enabled fraudstudent loans
Models

Mistral Releases Weights for Trillion-Parameter Mistral Large 4

Mistral has released the weights for Mistral Large 4 (ML4), a trillion-parameter open-source AI model. Businesses and governments can download and deploy the system on their own infrastructure. The release is positioned toward cybersecurity use cases.

AIHUBOct 30, 2026Score 72
Mistralopen-weight modelsAI in cybersecurity
Research Editor selected

SETI Researchers Use AI Agents to Operate Telescope and Analyze Live Signals

Breakthrough Listen and the SETI Institute are integrating AI agents into observations at the Allen Telescope Array, enabling telescope control, calibration and live signal analysis through natural-language requests. NVIDIA reports that the GPU-based pipeline handles 2.4 terabits per second and has demonstrated real-time AI detection of a pulsar in the Crab Nebula. The team is also evaluating open models, including NVIDIA Nemotron, for locally operated observatories.

Why it matters: The development offers a concrete example of AI agents controlling scientific instruments and analyzing high-throughput sensor data in real time.

NVIDIA BlogOct 30, 2026Score 87
SETI InstituteBreakthrough ListenNVIDIA
Research

Microsoft builds synthetic enterprise environments to evaluate Copilot at scale

Microsoft Research describes an Industry Evaluation Data Generation Platform that creates fully synthetic enterprise tenants for evaluating Microsoft Copilot without touching customer data. Environments are grounded in public sources such as O*NET occupational data, job postings, and SEC filings, and are paired with an automated query-generation pipeline across pharmaceutical, capital markets, and IT/software verticals. Microsoft reports a 39% gain on its composite synthetic-data quality metric versus its initial approach, with some industry environments scoring above 0.9 out of 1.0. The platform is published to an internal shared catalog and is positioned as the first standardized evaluation baseline for Copilot.

Microsoft ResearchOct 30, 2026Score 78
Microsoft Copilotsynthetic data evaluationenterprise AI
Research Editor selected

Epoch AI: Open-weight GLM 5.3 autonomously built working Firefox exploit chains in under 48 hours

Epoch AI researchers report that the open-weight model GLM 5.3 turned public Firefox security fixes into working exploit chains with zero human input, compromising six Firefox releases before updates reached users. Each chain reportedly cost roughly $1,000 in API fees. The runs did not trigger safeguards from any of the providers tested.

Why it matters: It offers a concrete, attributed measurement of autonomous offensive cyber capability in a widely available open-weight model, which matters to anyone deploying, evaluating or defending against such models.

the-decoderOct 30, 2026Score 88
GLM 5.3Epoch AIAI cyber capabilities
Models

Mistral releases trillion-parameter open-weight model despite dual-use warnings

Mistral has released a trillion-parameter AI model for free download, modification and local operation, placing it outside the developer's control. The developer itself warns the model could help attackers carry out real cyberattacks. The release is notable for making a very large open-weight model broadly available, together with an explicit security caveat.

AIHUBOct 30, 2026Score 72
Mistralopen-weight AI modelsAI cybersecurity risk