Skip to content

Archive / 2026-08-21

August 21, 2026

Blog

  • Agents Got Faster Than the Guardrails Meant to Stop Them

    OpenAI halted frontier training after its own models breached Hugging Face during a safety test. The rest of the industry's response — MCP patches, sandboxing tools, interim guidance — is still catching up, while enterprise survey data shows trust in automated oversight rising as it fails.

    TL;DROpenAI halted training after its own agents autonomously breached Hugging Face during a safety test, and the rest of the week's findings — leaky MCP servers, interim UK guidance, a sandboxing tool for coding agents, and enterprises trusting automated evals more even as 49% report evals missing production failures — show the same pattern: agent capability is outrunning the controls meant to contain it.

    Read the full post → https://engineerious.com/blog/2026-08-21-agents-got-faster-than-the-guardrails-meant-to-stop-them

News

View all news →

  1. Kagi added a setting for removing paywalled links from search results

    kagi.com27 days ago1252 ptsView detailsJoin discussion

  2. AI companies destroy physical books – let's scan rare books before it's too late

    annas-archive.pk28 days ago703 ptsView detailsJoin discussion

  3. I'm becoming AI-blind

    cymerys.com27 days ago487 ptsView detailsJoin discussion

  4. Rust Glancer: Rust LSP using 100x less RAM

    rust-glancer.github.io27 days ago420 ptsView detailsJoin discussion

  5. How a Texas student blew the whistle on a rogue AI hacking attempt

    reuters.com27 days ago188 ptsView detailsJoin discussion

  6. AI boosted homework scores, then exam scores dropped: Study

    canews24.online27 days ago165 ptsView detailsJoin discussion

  7. I ran Photoshop on a £0.60 computer chip

    pointinthecloud.com27 days ago151 ptsView detailsJoin discussion

  8. Initial focus for our partnership with Motorola is a regular non-folding device

    grapheneos.social27 days ago141 ptsView detailsJoin discussion

  9. Building an (almost) fully self-hosted, sandboxed, agentic software factory

    blog.jakesaunders.dev27 days ago119 ptsView detailsJoin discussion

  10. Radiation damage to Hubble has been 4.3 years out of phase with the Solar cycle

    arxiv.org27 days ago110 ptsView detailsJoin discussion

  11. Cancer-related mortality among US pilots and flight attendants

    cnn.com27 days ago99 ptsView detailsJoin discussion

  12. Show HN: OzBrain, a shared brain for knowledge between agents and your team

    ozbrain.com27 days ago92 ptsView detailsJoin discussion

  13. Kino: A high-performance Ractor web server for Ruby 4.0

    github.com27 days ago77 ptsView detailsJoin discussion

  14. Nvidia AVO scores 100% on the ARC-AGI-3 interactive reasoning benchmark

    twitter.com27 days ago70 ptsView detailsJoin discussion

  15. How I came to write that paper with Leslie Lamport

    lawrencecpaulson.github.io27 days ago58 ptsView detailsJoin discussion

  16. Seed: Minimal, self-modifying agent harness

    github.com28 days ago59 ptsView detailsJoin discussion

  17. 4 in 10 travellers visit tourist sites to capture social media content

    tuigroup.com27 days ago45 ptsView detailsJoin discussion

  18. Show HN: Proliferate- open-source, self-hostable Codex for any coding agent

    github.com27 days ago46 ptsView detailsJoin discussion

  19. LLMs are proof that Unix won

    bastian.rieck.me27 days ago41 ptsView detailsJoin discussion

  20. Rebuilding our Electron meeting-recording engine in Swift

    circleback.ai27 days ago36 ptsView detailsJoin discussion

  21. S&P 500 CEO median pay hits $17.3M, widening CEO-worker ratio to 312-to-1

    finance.yahoo.com27 days ago33 ptsView detailsJoin discussion

  22. Code Obfuscation via Local Mixing

    vitalik.eth.limo27 days ago34 ptsView detailsJoin discussion

  23. The disjointed Prague of Deus Ex: Mankind Divided

    killscreen.com27 days ago30 ptsView detailsJoin discussion

  24. Young Americans increasingly fear AI will take their jobs

    pewresearch.org28 days ago28 ptsView detailsJoin discussion

  25. Show HN: SteppeWM – a stacking IceWM-like Wayland compositor

    github.com27 days ago22 ptsView detailsJoin discussion

  26. Show HN: A desktop fly drawn to the scent of vibecode

    github.com27 days ago20 ptsView detailsJoin discussion

  27. Uber fined $966M in NL for automating driver suspensions

    reuters.com27 days ago20 ptsView detailsJoin discussion

  28. US Debt-to-GDP Ratio

    us-debt-clock.com27 days ago18 ptsView detailsJoin discussion

  29. LFM2.5-DSpark: Up to 3.2x Faster Inference from H100 to MacB

    liquid.ai27 days ago18 ptsView detailsJoin discussion

  30. The B-right/V R2 Operating System (2000)

    tronweb.super-nova.co.jp27 days ago18 ptsView detailsJoin discussion

  31. AI;DR or Don't be a meat proxy

    theaspiringnerd.com27 days ago18 ptsView detailsJoin discussion

  32. I'm Sick of Reading AI-Written Posts

    cyb3rops.medium.com27 days ago17 ptsView detailsJoin discussion

  33. DeepSeek-V4-Flash-Vision-Exp Is Now Live on the DeepSeek API Platform

    twitter.com28 days ago17 ptsView detailsJoin discussion

  34. Nvidia just showed that the harness, not the AI model, is now the real hero

    techcrunch.com27 days ago16 ptsView detailsJoin discussion

  35. Show HN: AgentSight – eBPF observability for AI agents, no code changes

    github.com27 days ago16 ptsView detailsJoin discussion

  36. We burned 11.7B tokens to find the best cyber AI model

    aikido.dev27 days ago15 ptsView detailsJoin discussion

  37. SCO Lawsuit agains Linux 20 years later

    lwn.net27 days ago13 ptsView detailsJoin discussion

  38. Nvidia AVO achieves 100% in ARC-AGI-3

    developer.nvidia.com27 days ago13 ptsView detailsJoin discussion

  39. Customs asked for his phone passcode. Now he faces a felony

    ajc.com27 days ago12 ptsView detailsJoin discussion

  40. The smell of rain: how Aussie scientists invented a new word in 1964

    csiro.au28 days ago12 ptsView detailsJoin discussion

  41. "No Disc, No Buy" Is Bigger Than Sony Expected

    gadgetreview.com27 days ago11 ptsView detailsJoin discussion

  42. Linus: "A debug session from hell, enormously helped by an AI"

    github.com27 days ago10 ptsView detailsJoin discussion

  43. Readers can't identify watermarked AI text

    seangoedecke.com27 days ago10 ptsView detailsJoin discussion

  44. Primary source

    An AI tool for prioritizing candidate biomarkers from wearable sensor data

    Generative AI

    research.google27 days agoView details

  45. Primary source

    From Atari to EVE Online: Building on 15 Years of AI Research in Games

    Google DeepMind partners with game studios to prototype breakthrough AI gameplay.

    deepmind.google27 days agoView details

  46. Primary source

    How mobility gives language models a deeper understanding of place

    Algorithms & Theory

    research.google27 days agoView details

  47. Best GPU Neoclouds 2026: CoreWeave, Nebius, Lambda, Crusoe, and Groq Ranked by Published Pricing and Contracted Power

    The five largest GPU neoclouds now run on very different models. CoreWeave and Nebius report to the SEC; Lambda and Crusoe are private and heading toward IPOs; Groq rebuilt itself as an inference cloud after licensing its LPU technology to NVIDIA. This comparison checks each provider's live rate card, Q2 2026 financia…

    marktechpost.com27 days agoView details

  48. Building Agentic Document Intelligence Pipelines: Creating Scientific Figures with AutoFigure

    This tutorial explores AutoFigure, a practical toolkit for generating professional scientific figures directly from text descriptions and research papers. We walk through setting up the environment, configuring an API-backed generation workflow, and converting complex document intelligence pipelines into publication-s…

    marktechpost.com27 days agoView details

  49. Anthropic Brings Claude Mythos 5 to Claude Security: Enterprise Teams Get Frontier Vulnerability Scanning Without Direct Model Access

    Anthropic has moved its most cyber-capable model into a product security teams can switch on themselves. Claude Security scans now run on Claude Mythos 5, in public beta for Claude Enterprise customers with no separate model add-on. The scan connects to a GitHub repository, traces data flows across files, and returns…

    marktechpost.com27 days agoView details

  50. The Unlikely Place at the Center of China’s AI Boom

    Cheap energy, abundant land, and proximity to Beijing have turned a city in Inner Mongolia into a crucial hub for data centers.

    wired.com27 days agoView details

  51. Over 1 million people have clicked LinkedIn’s AI slop button

    LinkedIn actually announced a "Seems like AI slop" button on July 30th, and the company says that a lot of people have already used it. According to a Thursday post from chief product officer Hari Srinivasan, "over a million people" have clicked on the button, which is accessible from the three dots menu on a post. Li…

    theverge.com27 days agoView details

  52. Major YouTube creators are facing backlash for accepting AI money

    A still from Sam Kolder’s video “AI is replacing me.” | Screenshot: The Verge, YouTube Over the past few days, a number of prominent filmmaking content creators including Matti Haapoja and Sam "Kold" Kolder have posted videos of themselves demonstrating what's possible with AI platform Higgsfield. The videos highlight…

    theverge.com27 days agoView details

  53. The Single English County Saying No to Palantir

    The UK government is facing calls to cancel a sprawling health care contract with Palantir. The region of Greater Manchester insists it can do a better job itself.

    wired.com28 days agoView details

Models

View all models →

  1. Audio8/Audio8-TTS-Preview-0.1b

    text-to-speech · transformers · safetensors · arktts

    huggingface.co28 days ago188 ptsView details

  2. EschaLabs/Qwen3.8-27B-Escha-W2

    text-generation · safetensors · qwen3_5 · qwen3

    huggingface.co28 days ago134 ptsView details

  3. outsourc-e/Qwen3.8-27B-Unleashed-GGUF

    text-generation · gguf · uncensored · abliterated

    huggingface.co27 days ago123 ptsView details

  4. bloomer010/Ling-3.0-flash-GGUF

    text-generation · llama.cpp · gguf · bailingmoe3

    huggingface.co28 days ago23 ptsView details

  5. magiccodingman/Qwen3.8-27B-MagicQuant-GGUF

    text-generation · safetensors · gguf · qwen3_5

    huggingface.co28 days ago3 ptsView details

  6. software-mansion/react-native-executorch-whisper-tiny.en

    automatic-speech-recognition · executorch · automatic-speech-recognition · en

    huggingface.co28 days ago1 ptsView details

  7. software-mansion/react-native-executorch-whisper-small

    automatic-speech-recognition · executorch · automatic-speech-recognition · license:apache-2.0

    huggingface.co28 days ago1 ptsView details

  8. software-mansion/react-native-executorch-whisper-tiny

    automatic-speech-recognition · executorch · automatic-speech-recognition · license:apache-2.0

    huggingface.co28 days ago1 ptsView details

  9. ModdiAdam/Wild_Krea-2-turbo_NSFW

    text-to-image · diffusers · gguf · text-to-image

    huggingface.co28 days ago1 ptsView details

  10. fpadovani/jpn-100mb-after-eng-baseline-ckpt500_seed455

    text-generation · transformers · safetensors · gpt2

    huggingface.co28 days agoView details

  11. fpadovani/eng-100mb-after-eng-baseline-ckpt500_seed455

    text-generation · transformers · safetensors · gpt2

    huggingface.co28 days agoView details

  12. software-mansion/react-native-executorch-whisper-base.en

    automatic-speech-recognition · executorch · automatic-speech-recognition · en

    huggingface.co28 days agoView details

  13. software-mansion/react-native-executorch-whisper-base

    automatic-speech-recognition · executorch · automatic-speech-recognition · license:apache-2.0

    huggingface.co28 days agoView details

  14. orpe42/MP_political_MIL_classifier

    transformers · tensorboard · safetensors

    huggingface.co28 days agoView details

  15. sullivan1502/base-zone-pretrain

    text-generation · transformers · safetensors · llama

    huggingface.co28 days agoView details

  16. mradermacher/DeepNemotron-3.5-Lightning-BF16-i1-GGUF

    text-generation · transformers · gguf · nemotron

    huggingface.co28 days agoView details

  17. RonnyMaurer255/MyAwesomeModel-TestRepo

    feature-extraction · transformers · pytorch · bert

    huggingface.co28 days agoView details

  18. XiAT/MyAwesomeModel-TestRepo

    feature-extraction · transformers · pytorch · bert

    huggingface.co28 days agoView details

Open source

View all open source →

  1. ggml-org/llama.cpp v0.2.0

    ## Overview This version marks the beginning of consistent semantic versioning for `llama.cpp`. The main goal is to make the job of downstream projects and users simpler when deciding which version of `llama.cpp`/`ggml` to distribute and use. The new version tags of the format `…

    github.com27 days agoView details

  2. langchain-ai/langchain langchain-perplexity==1.4.1

    Changes since langchain-perplexity==1.4.0 release(perplexity): 1.4.1 (#39826) fix(perplexity): include type="message" on Responses input items (#39774) fix(perplexity): preserve caller `extra_body` (#39203) chore: bump pillow from 12.2.0 to 12.3.0 in /libs/partners/perplexity (#…

    github.com27 days agoView details