AI news for builders and product teamsUpdated Nov 18, 2026, 03:01 UTC
AI news, straight from the source.
Releases, research, and ideas for developers and product teams. Updated every hour.
Across the feeds
Newest first
Databricks has been positioned in the Visionary quadrant of Gartner's Magic Quadrant for Analytics and Business Intelligence Platforms, marking its first appearance in the report.

GitHub has published research on a model for secret detection aimed at finding more leaked credentials with less noise, alongside related posts on Copilot customization and bug bounty methodology.

Linus Torvalds released an eight-patch Linux kernel series largely written with Anthropic's Claude Code, targeting a page-cache reference-counting bottleneck for multi-threaded small reads. His laptop tests showed eight-thread throughput rising from roughly 9.9-10.4 million to 30-33 million reads per second, but the branch remains unmerged.

Laplace Intelligence, a University of Michigan spinout based in Ann Arbor, is building automated red-teaming for robots and autonomous vehicles using AI agents inside interactive simulations. The profitable startup is competing in Startup Battlefield 200 at TechCrunch Disrupt in San Francisco on October 13-15.

Pluto Security has emerged from stealth with $42 million across an $8.5M seed round led by TLV Partners and a $34M Series A led by Jump Capital. The startup sells software that watches how employees use AI tools and agents, and says about 60% of the risks it detects come from vibe-coded apps built by non-developers.

AWS published a sample repository showing two ways to activate AWS Marketplace Private Pricing Offer entitlements for Amazon Bedrock in AWS License Manager: a one-time script and an event-driven CDK stack. Organization-level grants are auto-accepted in AWS Organizations but stay Disabled until activated, so negotiated pricing does not apply until someone acts.

Michael I. Jordan, professor emeritus at UC Berkeley and now a full-time researcher at Inria in Paris, has sharply criticized AI existential-risk warnings, calling them science fiction, dismissing researchers' resignations as a "grand moralizing performance" and companies' calls for regulation as "theater."

Liquid AI released LFM2.5-24B-A2B, a 24B-parameter reasoning agent model with only 2B active parameters per token, designed to run agents locally on laptops and single GPUs. It scores 48.09 on LiveClaw versus 38.19 for Qwen3.6-35B-A3B and decodes 184 tok/s on an Apple M5 Max.

Meta is bringing its Your Algorithm feed customization feature to Facebook, letting users add and manage topics for their feed and Reels after it launched on Instagram earlier this year.

Google is expanding Play Store AI features, including a broader Ask Play search and Game Finder with a Gamer DNA profile, while adding a Personal Library that lets AI Studio users privately host vibe-coded apps without a developer account.

Google AI Studio now lets anyone with a standard Google account build personal Android apps from text prompts at no cost, installing them on their own device rather than publishing to the public Play Store. Sharing with family and friends is slated for the coming months.

Google Play now lets users in the US aged 18 and over run personal, AI-built Android apps on their own devices without publishing them publicly. The apps stay private, are not searchable, and must still follow Play policies; sharing with trusted friends and a path to public release are planned.

Google now lets US users launch personally AI-coded apps directly through the Play Store using a standard Google account, without connecting a phone to a computer. The apps stay private and separate from the public Play Store catalog, with selective sharing coming soon.

Orange Business built Live Intelligence, a governed AI platform on LangChain and LangSmith, deployed to 120,000 employees and adopted by 150+ customers in France and Europe. It uses a three-layer evaluation stack and agent and MCP registries for cross-organization governance.

Satlyt, a Sunnyvale- and Nairobi-based startup founded by former Google and SpaceX staffer Rama Afullo, raised an $8 million seed round led by Non Sibi Ventures to build open software for running AI on satellites. Its software launches Thursday on a SpaceX rocket aboard a TakeMe2Space spacecraft with NASA among three customers.

LangChain reports that open-weight models GLM-5 (z.ai) and MiniMax M2.7 now score similarly to closed frontier models on core agent tasks such as file operations, tool use and instruction following, at a fraction of the cost and latency, making them viable for production agents in Deep Agents.

LangChain migrated inline documentation code samples into standalone, CI-tested files using the Deep Agents CLI driven by a reusable skill definition. The skill lives at .agents/skills/docs-code-samples/SKILL.md in the docs repo and handles file creation, snippet markup, linting and test runs.

LangChain released Open SWE, an MIT-licensed open-source framework for building internal coding agents, built on Deep Agents and LangGraph and modeled on publicly described systems at Stripe, Ramp, and Coinbase. It ships with pluggable cloud sandboxes, curated tools, Slack, Linear and GitHub triggers, subagent orchestration and deterministic middleware.

LangChain detailed how its open source Deep Agents SDK keeps long-running agents inside a model's context window using three compression steps: offloading large tool results at 20,000 tokens, evicting old file-write arguments at 85% of the window, and summarizing the conversation history when nothing else can be offloaded.

LangChain launched Deep Agents Deploy in beta, a single command that packages an open source, model agnostic agent harness with its own LangSmith Deployment server. It exposes agents via MCP, A2A and Agent Protocol, supports multiple model and sandbox providers, and lets teams self-host so agent memory stays in their own databases.