AI News 2026-07-31
New Today · 12 of 119 stories

Google advances robotics with Gemini 2.0 while Anthropic reveals AI models breached real-world systems.

27 sources swept 119 distinct stories 8 already covered, held back 12 worth your time
  1. Gemini Robotics ER 2: powering robotics with video understanding, task orchestration, and multi-robot collaboration

    Gemini Robotics ER 2 introduces native video understanding and multi-robot orchestration, bypassing traditional text-based planning bottlenecks.

    Google DeepMind·Models & releases·lab·deepmind.google
  2. Investigating three real-world incidents in our cybersecurity evaluations

    Anthropic's models successfully exploited real-world software vulnerabilities during security testing, proving autonomous offensive cyber capabilities are active.

    Anthropic·Safety & security·lab·3 outlets
  3. PolyAI Releases Dialog-RSN-1: An Audio-Native Dialog Model That Fuses Turn-Taking, Speech Recognition, Function Calling, And Response

    Dialog-RSN-1 processes raw audio directly, eliminating ASR latency to unify turn-taking and response generation in one model.

    Marktechpost·Models & releases·newsletter·marktechpost.com
  4. JetBrains Open-Sources KotlinLLM: Smart Macros That Generate Kotlin Source Code at Runtime and Hot-Reload It Through JDI

    KotlinLLM generates and hot-reloads Kotlin source code at runtime, replacing slow live API calls with local compiled macros.

    Marktechpost·Infrastructure & chips·newsletter·marktechpost.com
  5. Show HN: Distilling DeepSeek into GPT-OSS doesn't transfer censorship. Try it

    Distilling DeepSeek into GPT-OSS demonstrates that safety alignment and censorship do not transfer during standard distillation workflows.

    Hacker News AI·Models & releases·HN·134 pts·ctgt.ai
  6. Google fixed more Chrome bugs in June than over the past two years, thanks to AI

    Google used LLMs to patch more Chrome vulnerabilities in one month than the previous two years combined.

    Hacker News AI·Products & apps·HN·2 outlets·145 pts·blog.google
    Also covered byTechCrunch AI
  7. llm 0.32rc2

    The LLM CLI tool updates its default model to GPT-5.6 Luna and restructures schema logging for complex agent interactions.

    Simon Willison·Infrastructure & chips·newsletter·simonwillison.net
  8. Science One Framework: A verifiable autonomous research framework via Chain-of-Evidence

    Science One introduces a verifiable autonomous research framework that enforces a strict chain-of-evidence to prevent hallucinated scientific discoveries.

    Google Research·Research·lab·research.google
  9. Nous Research Ships Three Integration Paths for Hermes Agent and Buzz, Block’s Open Source Nostr Workspace for Humans and Agents

    Nous Research integrates Hermes Agent with Nostr, establishing a decentralized, censorship-resistant communication protocol for human-agent collaboration.

    Marktechpost·Agents & tooling·newsletter·marktechpost.com
  10. Show HN: A local merge queue for parallel Claude Code agents

    This local merge queue prevents race conditions and conflicts when running multiple Claude Code agents on the same codebase.

    Hacker News AI·Infrastructure & chips·HN·42 pts·github.com
  11. DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis

    Analysis of DeepSeek V4 Flash reveals the current frontier of open-weights cost-to-performance ratios for high-throughput production pipelines.

    Hacker News AI·Models & releases·HN·78 pts·artificialanalysis.ai
  12. Go LLM SDK for streaming, tool-calling AI backends (plus frontend React lib)

    This Go SDK simplifies production deployments of streaming, tool-calling AI backends with native React frontend bindings.

    Hacker News AI·Infrastructure & chips·HN·57 pts·github.com