0x27 Cyberhades Podcast: Xenix, Turing y "Los pollito mi compare Migue", supera este cóctel IA

Hoy venimos con un episodio bien cargado: hablamos de Xenix, el Unix de Microsoft antes de MS-DOS y Windows; del salto de los modelos de IA a la ciberseguridad real; de Mythos, GPT-5.5-Cyber, Patch the Planet y cómo la IA ya no solo encuentra vulnerabilidades, sino que también empieza a parchearlas. También comentamos AutoJack, agentes locales, MCP, WebSockets en localhost, routers domésticos convertidos en infraestructura criminal y el problema de distinguir entre una vulnerabilidad y algo realmente explotable.

Además, nos vamos por el lado más geek: Alan Turing y su dispositivo portátil de cifrado de voz, el Manchester Baby, Linux moderno corriendo en un Motorola 68008, Ubuntu controlado por lenguaje natural, Qwable v1, Role Confusion, Quake, el Tatung Einstein y hasta una RTX 3060 refrigerada con una máquina de hielo. También hablamos de nuestro libro Hacking IA, que ha llegado al número uno en Amazon en la categoría de Internet y Web en inglés, y de cómo la IA se está metiendo ya en programación, medicina, derecho, cloud, hardware y prácticamente cualquier tema tecnológico.

Enlaces de este episodio:

• Startup Claims Breakthrough in Major LLM Bottleneck

  https://www.technologyreview.com/2026/06/19/1139313/a-startup-claims-it-broke-through-a-bottleneck-thats-holding-back-llms/

• John Jumper Leaves DeepMind for Anthropic

  https://economictimes.indiatimes.com/tech/artificial-intelligence/nobel-prize-winner-john-jumper-quits-google-deepmind-after-nearly-a-decade-set-to-join-anthropic/articleshow/131866136.cms

• DeepMind’s Roadmap for Controlling AI When Alignment Fails

  https://www.techtimes.com/articles/318758/20260620/google-deepmind-ai-control-roadmap-when-alignment-fails-defense-depth-takes-over.htm

• OpenAI’s New Life-Sciences Benchmark Exposes Model Weaknesses

  https://science.slashdot.org/story/26/06/20/202204/openai-announces-benchmarks-for-ai-life-sciences-research-its-best-model-failed-639-of-the-test

• Anthropic’s Claude Suffers Major Service Outage

  https://cybersecuritynews.com/anthropic-claude-ai-outage/

• Anthropic’s Mythos Model Sparks Discussion

  https://cybersecuritynews.com/anthropics-mythos-ai-model/

• A 13-Word Reddit Comment Tricks ChatGPT and Gemini

  https://cybersecuritynews.com/13-word-reddit-comment-chatgpt-and-gemini/

• GPT-5.6 Rumors: Alignment Fixes and Massive Context Windows

  https://www.techtimes.com/articles/318799/20260621/gpt-56-launch-window-starts-monday-alignment-fix-15m-token-context-inside.htm

• Google’s AMIE Outperforms Physicians in Simulated Care Tasks

  https://www.news-medical.net/news/20260621/Googlee28099s-AMIE-beats-doctors-on-key-simulated-disease-management-tasks.aspx

• GPT-5.5 and the Future of AI-Powered Cybersecurity

  https://cybersecuritynews.com/gpt-5-5-cyber/

• AI Law Firm Wins UK Court Case for the First Time

  https://yro.slashdot.org/story/26/06/22/1657204/ai-law-firm-wins-uk-court-case-for-first-time

• Canonical Wants You to Talk to Ubuntu Instead of Typing

  https://news.slashdot.org/story/26/06/21/0234252/canonicals-upcoming-ai-tool-talk-to-ubuntu-instead-of-typing

• Qwable-v1: A New Open Model Arrives on Hugging Face

  https://huggingface.co/lordx64/Qwable-v1

• AutoJack Targets Modern Vehicle Ecosystems

  https://cybersecuritynews.com/autojack/

• Millions of Home Devices Secretly Fuel Cyberattacks

  https://it.slashdot.org/story/26/06/20/2321236/how-millions-of-digital-home-devices-are-secretly-powering-cyberattacks

• New Unpatchable Exploit Impacts Apple A12 and A13 Devices

  https://it.slashdot.org/story/26/06/19/1723242/new-unpatchable-exploit-targets-apple-devices-with-a12-and-a13-chips

• Hackers Abuse GitHub to Distribute Malicious Scripts

  https://cybersecuritynews.com/hackers-github-malicious-script/

• AI-Powered iOS Apps Expose LLM Credentials

  https://cybersecuritynews.com/ai-ios-apps-leaking-llm-credentials/

• Vulnerability Found in Apple Beats Studio Buds

  https://cybersecuritynews.com/apple-beats-studio-buds-vulnerability/

• SquidBleed Threatens Enterprise Proxy Deployments

  https://cybersecuritynews.com/squidbleed-vulnerability/

• Tata Electronics Reports Data Breach

  https://cybersecuritynews.com/tata-electronics-data-breach/

• Patch the Planet: Trail of Bits Launches Open-Source Security Initiative

  https://blog.trailofbits.com/2026/06/22/introducing-patch-the-planet/

• Role Confusion: New Attack Class Against AI Agents

  https://role-confusion.github.io/

• GitHub Experiences Service Disruption

  https://x.com/github/status/2069025193620648259?s=20

• Running Modern Linux on a Motorola 68008

  https://hackaday.com/2026/06/19/running-modern-linux-on-a-68008/

• Remembering Alan Turing’s Portable Voice Encryption Device

  https://it.slashdot.org/story/26/06/20/059229/remembering-when-alan-turing-developed-a-portable-voice-encryption-device

• Sound Waves Could Reduce Espresso Brewing Energy by 75%

  https://developers.slashdot.org/story/26/06/19/2159245/using-sound-waves-to-make-espresso-could-cut-coffee-brewing-energy-use-by-75

• Rolls-Royce Wins Swedish Small Nuclear Reactor Contract

  https://hardware.slashdot.org/story/26/06/18/2335235/rolls-royce-secures-deal-to-build-small-nuclear-reactors-for-sweden

• Modder Turns an Ice Machine into an RTX 3060 Cooler

  https://www.tomshardware.com/pc-components/gpus/ingenious-modder-converts-countertop-ice-machine-into-an-rtx-3060-gpu-cooler-with-a-beer-fridge-thermostat-mod-reduces-temps-by-up-to-62-percent-in-games-cyberpunk-2077-runs-at-22-c

• Ubisoft Co-Founder Claude Guillemot Dies in Plane Crash

  https://www.tomshardware.com/video-games/ubisoft-co-founder-claude-guillemot-dies-in-plane-crash-french-publisher-established-in-1986-became-one-of-the-biggest-entertainment-companies-in-the-world

• Is Tesla Preparing to Sell Modular AI Datacenter Hardware?

  https://hardware.slashdot.org/story/26/06/21/2145245/is-tesla-planning-to-sell-modular-ai-data-center-hardware

• Tutor Jailed for Taking Online Exams for 124 Students

  https://news.slashdot.org/story/26/06/22/0255256/tutor-who-took-online-tests-for-124-students-jailed-for-three-years

Como siempre puedes ver o escuchar el episodio en:

• YouTube
• Spotify