AI news for builders and product teamsUpdated Nov 18, 2026, 10:01 UTC
AI news, straight from the source.
Releases, research, and ideas for developers and product teams. Updated every hour.
Across the feeds
Newest first
Engineering teams are rethinking code review as AI-generated code floods their pipelines, according to a Sonar survey of over 1,100 developers who estimated AI contributed 42 percent of code added to shared codebases while 96 percent did not fully trust its output.

Zhipu's GLM-5.3 moved from an MIT license to a custom one requiring inference and fine-tuning providers with over $10 billion in 12-month revenue to pass a Z.AI security review before commercial use. The brief also covers Motif-3, dots3-note-prev, Qwen3.8-Flash-Next, Hy4-preview and NVIDIA's Nemotron update.

Google DeepMind announced the AlphaGenome Atlas on 8 September, an online repository of precomputed AlphaGenome predictions covering all 9 billion possible single-letter changes to a reference human genome. It is freely available for noncommercial research, with potential commercial licensing.

Astra, described as the start of the GPT-6 family, has been released and is said to top ARC-AGI-3 and Zapier's AutomationBench while costing the same as Fable 5.1. Commentators describe it as powerful but inconsistent, and OpenAI says it has met its automated research intern goal.

OpenAI has released GPT-6 Astra, a new OpenAI agent message board has been discovered, and Anthropic launched Claude Fable 5.1, which it says is up to 45 percent cheaper for agentic work.

GitHub introduced Project HydraFusion, a research preview available on all GitHub Copilot plans through /experimental in GitHub Copilot CLI. It orchestrates models across providers at runtime, selecting among Single, Cascade, and Critique execution patterns based on task needs.

Ben's Bites built a map-based site from 107 million rows of UK council spending data, assembled largely by Codex agents across 656 subagents and 8.2 billion tokens, with 319 of 339 targeted councils collected or parked.

LangChain moved MCP support into the core langchain package as langchain.mcp, built on FastMCP for the 2026-07-28 spec, adding elicitation through LangGraph interrupts and client-side tool list caching.

GitHub published a beginner-focused post describing how its Copilot app lets developers run multiple agent sessions at once, each on its own Git worktree. The post walks through running three tasks in parallel in an example repository and includes a call to get started with the app.

Hugging Face announced NeoMME, a family of 260M and 800M multilingual multimodal encoders that process text tokens and raw image patches in a single bidirectional Transformer trained from scratch. Checkpoints are released under the Apache 2.0 license, with a fine-tuned NeoMME-Retriever for visual document retrieval.

Anthropic released Claude Fable 5.1, which Ben's Bites describes as faster and easier to talk to, with a new system prompt restricting song lyrics and copyrighted logos. Anthropic also cut the cost of caching inputs for Fable by 75%, making API usage roughly 25% cheaper.

Hugging Face has published funes, an open-source local memory layer for coding agents such as Claude Code, Codex, pi, and Hermes. It indexes existing session traces, supports recall and ask commands, and can optionally sync memory to a private Hugging Face dataset you own.

A Hugging Face blog post describes an open reproduction of a project that trains a coding model to paint watercolours by writing p5.brush JavaScript, using TRL and OpenEnv with three reward mixes. The blog reports trained models and datasets published openly on the Hub.

A Hugging Face guide describes fine-tuning LiquidAI's LFM2.5-350M with GRPO via TRL, improving IFStruct structured-output scores from 22.6% to 29.7% in about 100 steps and roughly 500 samples. The run fits a free-tier Colab or Kaggle GPU, and evaluation ran locally through llama.cpp on a MacBook.

A MIT News feature profiles three IBM researchers—Srinivasan Arunachalam, Zhang-Wei Hong PhD ’25, and Irene Ko PhD ’24—who credit the MIT-IBM Computing Research Lab with helping translate theoretical work into industry applications in quantum machine learning, reinforcement learning, and trustworthy AI.

Google launched the Fairwind Program, a limited access initiative giving governments and trusted partners access to Gemini 3.8 Flash Cyber and the CodeMender harness to autonomously find and fix vulnerabilities. More than 650 partners participate globally, with strict operational standards required.

MIT and Motional researchers developed the Concept-Wrapper Network (CW-Net), a method that translates a self-driving car's deep learning decisions into understandable concepts without altering driving performance. In private-track tests and simulation studies, CW-Net explanations helped safety drivers and nonexpert users better predict vehicle behavior. The research appears in Nature.

An IEEE Spectrum essay by systems engineer Richard Mitchell argues that letting AI absorb entry-level engineering work erodes the apprenticeship pipeline that builds senior expertise, citing aviation and nuclear control-room precedents. He proposes deliberate "manual gates" that keep humans practicing skills they will need when automation fails.

Researchers introduced BenchMIRT, a multidimensional Item Response Theory method for auditing LLM benchmarks at the level of individual prompts. Trained on results from 100 LLMs across 16 benchmarks and more than 34K questions, it independently recovered safety and general reasoning as dominant dimensions.

Walter Torous, senior lecturer at MIT's Department of Urban Studies and Planning and MIT Sloan, was named executive director of the MIT Center for Real Estate effective July 1, 2026. Justin Steil will serve as faculty chair of the Academic Curriculum Council.