FeedBag v1.2.4
近 3 天新闻 · 共 14 条
Java News Roundup: Lazy Constants, Spring for Apache Pulsar, GlassFish, Hibernate, TomEE, Seed4J
InfoQ · 1 天前
This week's Java roundup for November 9th, 2026, features news highlighting: Lazy Constants proposed to target for JDK 28; TornadoVM jitLLM 3.0; point releases of Seed4J; second release candidates of Spring AI and Spring Boot; first release candidates of Hibernate ORM and Hibernate Reactive; fourth milestone release of GlassFish; and the second milestone release of Apache TomEE. By Michael Redlich
Uber’s Legal AI Agent Cuts Contract Review Time by 20% with Rules and RAG
InfoQ · 1 天前
Uber’s Legal Redlining Agent reviews contract changes in Microsoft Word using RAG, historical lawyer feedback, and deterministic policy rules. Built with LangGraph, OpenSearch, and Redis, the system reports more than 20% lower average contract review time and 91% accuracy in AI-generated decisions. Uber is exploring agentic harnesses, skills, legal ontologies, and knowledge graphs. By Leela Kumili
Airbnb Introduces CASTLE for Training Natural Language Search Models with Synthetic Data
InfoQ · 1 天前
Airbnb researchers have introduced CASTLE, a framework designed to address the cold-start problem in natural language search. The approach uses de-identified booking sessions and large language models (LLMs) to jointly generate synthetic search queries and relevance labels, enabling model training before sufficient real search traffic is available. By Robert Krzaczyński
Article: Governed Autonomous Intelligence for Mission-Critical Utility
InfoQ · 1 天前
The article describes an architecture for agentic AI in regulated utility operations. Agents reason and propose actions, but a deterministic policy gate and a governed tool gateway control execution, with human approval where required. It covers an AWS reference implementation, grounding in runbooks, governed prediction, outcome validation, audit evidence, and phased autonomy. By Purnang Borah
Podcast: Getting CRA Compliant Hardens the Code and Gets Your Company Closer to Digital Sovereignty
InfoQ · 1 天前
In this episode, Christopher "CRob" Robinson, CTO of OpenSSF, Akrites and Chief Security Architect of the Linux Foundation, explores the state of software supply chain security amid multiple security incidents and explains how tools like Scorecard, the Security Baseline framework, and SLSA help harden software. By Christopher Robinson
AWS Open-Sources Emulator for Local Testing of 200+ Services by Coding Agents
InfoQ · 1 天前
AWS Emulator is a generally available local HTTPS endpoint, generated from public Smithy API models, that lets developers and coding agents test against more than 200 AWS services without an account, with published fidelity scores and documented limits around IAM, regions, and in-memory state. By Daniel Dominguez
GitHub Rebuilds the Copilot App's Diff Surface to Open Million-Line Pull Requests Without Jank
InfoQ · 1 天前
GitHub engineers rebuilt the diff view in the Copilot app to handle a pull request with 2,200 files, over a million changed lines and 400 inline comments. The design separates fixed-height code rows from variable-height comment blocks, which are measured in batches near the viewport. Performance is checked through CI budgets and automated app runs. By Daniel Curtis
AWS Nitro Sandboxes Put Each EKS Pod in Its Own MicroVM with One RuntimeClass Field
InfoQ · 1 天前
AWS has launched Nitro Sandboxes, running each opted-in EKS pod, ECS task or EC2 workload in its own Firecracker microVM with a separate kernel, enabled by one runtimeClassName field at no extra charge. A confidential mode adds attestation-gated key release. The documentation lists the trade-offs: no GPU passthrough, at most five sandboxes per vCPU, and half the compute on SMT instances. By Steef-Jan Wiggers
Alpine Linux Users Urged to Patch OpenSSH Authentication Flaw
InfoQ · 1 天前
A recently disclosed vulnerability in OpenSSH lets an unauthenticated attacker log into accounts that should have no usable password. The issue, named *sshole and tracked as CVE-2026-123158, affects Linux systems running OpenSSH with the musl C library and which don't use PAM. Security researchers now report that attackers are exploiting it in the wild. By Matt Saunders
Uno Platform 7.0 Adds WebGPU Backend, Apple TV and Reactor Support, and Retires Native Rendering
InfoQ · 1 天前
Uno Platform 7.0 unifies rendering on SkiaSharp 4 with an opt-in WebGPU backend, adds Apple TV as a target and supports .NET 11 from day zero. It also runs third-party .NET MAUI controls and Microsoft.UI.Reactor on all targets, and removes native rendering and Mac Catalyst. By Edin Kapić
Uber's Secure AI Framework Governs Autonomous AI Agents across Their Lifecycle
InfoQ · 2 天前
In a recent article, Uber Engineering highlights the limitations of traditional identity and API controls in securing AI agents that dynamically reason over external content and take real-world actions. Uber developed a framework to determine whether an agent's action is appropriate given the user's request, the agent's intended purpose, its permissions, and the context of the interaction. By Sergio De Simone
Anthropic Outlines Its AI-Powered Code Review Process for Claude Code
InfoQ · 2 天前
Anthropic has outlined how the Claude Code team handles code reviews as AI agents increasingly contribute to its codebase. The approach relies on multiple layers of automated verification, AI-powered reviews, and risk-based human approval, aiming to reduce manual review time without increasing regressions or production incidents. By Robert Krzaczyński
Cloudflare Introduces Private Origins for Application Services
InfoQ · 2 天前
Cloudflare recently made Application Services for Private Origins generally available, allowing public traffic to reach applications hosted on private networks without exposing their IP addresses or opening inbound firewall ports. Private-to-private connectivity is not yet supported. By Renato Losio
Cohere Finding Shows Self-Hosted Open-Weight Models Still Need Egress Controls
InfoQ · 2 天前
Cohere recently reported that one of Alibaba’s newest Qwen3.8 models attempted to upload a probe file to a storage container in Beijing while explicitly stating in its reasoning trace that no tools would be triggered. While there is no evidence of intentional backdooring, the report is a reminder that, even for self-hosted agents based on open-weight models, egress risks exist and need mitigation. By Bruno Couriol

Airbnb Introduces CASTLE for Training Natural Language Search Models with Synthetic Data

InfoQ · 1 天前
Airbnb researchers have introduced CASTLE, a framework designed to address the cold-start problem in natural language search. The approach uses de-identified booking sessions and large language models (LLMs) to jointly generate synthetic search queries and relevance labels, enabling model training before sufficient real search traffic is available. By Robert Krzaczyński
查看原文 ↗