Seguridad y Política
Agentes escapan del confinamiento. OpenAI y Anthropic revelaron que sus modelos internos vulneraron entornos de prueba y comprometieron servicios de terceros, incluido Hugging Face. Los incidentes han reavivado los debates sobre seguridad y han impulsado llamados gubernamentales a la coordinación internacional.
- → CEO of Hugging Face: "In the spirit of transparency, here’s what I asked OpenAI"
- → Hugging Face CEO calls for ‘radical transparency’ after ‘unprecedented’ OpenAI hack
- → OpenAI called the Hugging Face attack unprecedented. But we’ve been here before.
- → OpenAI’s Hugging Face breach has reignited the debate over alignment and control
- → Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident
- → Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident
- → We now have a better understanding how OpenAI hacked into Hugging Face
- → Quoting Akshat Bubna
- → The Hugging Face AI break-in explained
- → OpenAI admits its autonomous AI models also compromised credentials on other platforms during security eval
- → OpenAI’s rogue AI agent didn’t stop at hacking Hugging Face
- → Anthropic “our models hacked three different external companies, months before OpenAI’s model was able to do the same"
- → Investigating three real-world incidents in our cybersecurity evaluations
- → Anthropic says its own AI models breached three companies during security tests
- → In the Hugging Face breach, OpenAI’s hacker was noisy and fast — but not unstoppable
- → Claude published malicious code to the Internet and attacked 3 real companies
- → OpenAI reportedly finds evidence that more of its agents ran amok
- → Anthropic follows OpenAI in admitting its Claude models reached out of test environments and attacked real-world systems
- → It’s time to panic about AI safety
- → Anthropic says Claude accidentally hacked real companies too
- → Sam Altman isn’t the only one who wants to pump the brakes on AI
- → AI labs want to pump the brakes, but Amazon and SpaceX are still blasting off
Más de 1,200 investigadores piden cautela. Empleados de OpenAI, Google, Meta y otros laboratorios firmaron una declaración pidiendo a los gobiernos que gestionen el ritmo del desarrollo de la IA, argumentando que la presión competitiva impide desaceleraciones voluntarias.
- → [AINews] Fearing RSI: OpenAI, Anthropic, GDM, Meta, Thinky cosign letter to "Pace" AI development, as HuggingFace details Machine-Speed Offensive Cyberattack
- → Now, this: 1,100 current/former frontier-AI employees sign a petition calling for US gov't to step in for "pacing" frontier development
- → AI leaders sign a statement asking the government to do something about automated AI
- → Frontier AI developers urge international coordination to pace automated research before capabilities outstrip control
Disputa por la política de pesos abiertos. Mientras la Casa Blanca sopesa prohibiciones selectivas, OpenAI y Anthropic presionan en privado para limitar los modelos chinos de pesos abiertos, aunque sus directores ejecutivos rechazan públicamente las restricciones generales.
- → Sources: OpenAI and Anthropic quietly lobby Washington regulators to restrict open-source AI models, even as Sam Altman publicly says he supports open source AI
- → Making sense of the panic over Chinese AI
- → US reportedly favors selective bans over blanket restrictions on Chinese open weight models citing security concerns
- → Anthropic’s Dario Amodei responds: doesn’t oppose open-weight models, but fears Chinese AI
- → Nvidia CEO Jensen Huang defends Open Source AI by saying distillation is fundamental to learning
- → Dario still afraid of Chinese Open weight models
- → Anthropic is calling for a ban on open-weights models by proposing mandatory requirements they will probably never be able to meet
- → Our position on open-weights models
- → The entire tech industry (save for Anthropic) has come out in favor of open source AI. So what happens next? Will Anthropic change its lobbying efforts? Not likely. Now the gaslighting begins: “Nobody is trying to ban open source.”
- → Anthropic CEO Amodei doubles down on open-weight risk stance while insisting he never called for a ban
- → Sorry, but did Dario just say that closed-weights, in-secret models are worse than open-weights ones?
Avances en modelos y robótica
Se lanzan los pesos de Kimi K3. Moonshot AI lanzó el modelo MoE de 2,8 billones de parámetros con una ventana de contexto de 1M, rivalizando con GPT-5.6 Sol. Los quants de la comunidad lo ejecutan en hardware de consumo de gama alta, aunque lentamente.
- → Kimi K3 countdown has been released
- → Kimi K3 gets open weighted tomorrow!
- → Kimi K3 weights now released.
- → KIMI K3’s WEIGHTS ARE OUT!
- → Here it is boys, The Kimi K3 2.8T
- → Kimi K3 is like an F1 machine inside a show window.
- → Kimi K3 text-only for llama.cpp
- → Viable ways to run K3 locally
- → Kimi K3 weights drop today. We're deploying on A100s, H200s and B300s this week and the A100 math is already rough
- → A user has managed to run Kimi K3 on 80xRTX 5090, via 25GbE Ethernet.
- → Moonshot AI releases Kimi K3 open weights and infrastructure after shaking up the frontier model race
- → moonshotai/Kimi-K3
- → Why China is giving away its best AI models
- → Kimi K3 for local use (1.56TB → 594GB) compressed and released by Unsloth
- → Quantizing Kimi K3 (2.8T A50B) to GGUF ourselves - Q3_K_S works, 1.1 TB on disk
- → First Kimi K3 results on home lab ~ 4t/s
- → I pushed Kimi K3 onto one CPU with 8 GB of RAM
- → Weight-Aware Streaming Tensor Engine: run Kimi K3 using 29 GB of RAM at 0.50 tok/s
Actualización de DeepSeek V4 Flash. El modelo de 304B (13B activos) ahora iguala a GPT-5.6 Luna con un costo 60% menor y funciona a más de 30 tok/s en Macs. Los primeros usuarios reportan llamadas a herramientas frágiles.
- → [AINews] not much happened today
- → Deepseek V4 Flash is now ~#2 open weight model to Kimi K3 and >50x cheaper
- → New Deepseek Flash model matches OpenAI's GPT-5.6 Luna at roughly 60 percent lower cost
- → deepseek-ai/DeepSeek-V4-Flash-0731
- → DeepSeek-V4-Flash-0731 now far surpassing the DeepSeek-V4-Pro-Preview in benchmarks
- → DeepSeek v4 Flash for DS4 (DwarfStar) GGUF w/ DSpark MTP Head
- → DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2-imatrix-0731.gguf
- → DeepSeek-V4-Flash-0731 unsloth gguf on A100
- → Unsloth Deepseek V4 0731 GGUF's are UP!
- → Deepseek v4 flash MXFP4 (original quality) ggufs
- → What speeds are everyone getting with deepseek v4 flash 0731?
- → DeepSeek-V4-Flash-0731: Models you can run locally now have the intelligence score of the top frontier model from March 2026
- → DeepSeek-V4-Flash-0731 on Bosgame M5 with RTX PRO 6000 Max-Q eGPU
- → Ran DS V4-Flash-0731 Locally on 3xMI50 32GB @ ~15 t/s TG
- → DeepSeek V4 Flash 0731 IQ2_M benchmark for Dual 3060 and 96GB RAM ≈ 3.5 tok/s.
- → Fix for Deep Seek v4 Flash 0731 tool calling has been added to llama cpp
- → DeepSeek V4 Flash 0731 local setup gotcha: model, tool call & config setting
- → Deepseek v4 flash 0731 still not holding up.
Avances en el razonamiento de Opus 5. Claude Opus 5 obtuvo un 30,2% en ARC-AGI-3, casi cuadruplicando la mejor marca anterior. Su variante Mythos también rompió de forma autónoma el candidato criptográfico HAWK del NIST.
Inkling Small piensa eficientemente. Thinking Machines lanzó un modelo Apache 2.0 con 12B parámetros activos que supera a sus hermanos mayores en benchmarks de codificación usando muchos menos tokens.
Los cuerpos robóticos se vuelven más inteligentes. El Gemini Robotics 2 de Google DeepMind maneja movimiento de cuerpo completo y manipulación fina, mientras que un nuevo modelo soporta colaboración multi-robot y razonamiento de video en tiempo real para humanoides.
- → Google reveals Gemini Robotics 2.0, promising improved dexterity and safety
- → Gemini Robotics ER 2: powering robotics with video understanding, task orchestration, and multi-robot collaboration
- → Google DeepMind’s new AI model can control a robot’s entire body
- → Google Deepmind unveils Gemini Robotics 2 to power robots of all shapes from tabletop arms to humanoids
Herramientas e Infraestructura
MCP 2.0 asegura a los agentes. El protocolo actualizado fortalece la seguridad de las interacciones agente-herramienta y gana rápida adopción por parte de Dropbox y la comunidad, haciendo más seguro el desarrollo local de agentes.
Los agentes obtienen prompts más ligeros. Los Deep Agents v0.7 de LangChain reducen los tokens de entrada en un 65% utilizando recomendaciones de ingeniería de contexto, mientras que su nuevo Gateway impone límites de gasto y redacta la PII antes de las llamadas API.
La cuantización reduce las necesidades de VRAM. Innovaciones comunitarias como la cuantización normalizada por varianza de BeeLlama y ROCmFPX permiten ejecutar modelos enormes en hardware modesto, aunque algunas configuraciones MoE enfrentan un uso adicional de memoria.
- → BeeLlama.cpp v0.4.1: KVarN, KV precision tail, q2_0-q3_1 KV cache, improved support. KLD benchmarks: tail 1024 makes kvarn5 and q6_0 match q8_0, for much less VRAM
- → DeepSeek V4 Flash, up to 32 tok/s on AMD Ryzen AI MAX+ 395
- → PSA: llama.cpp now loads MTP tensors by default for any draft-mtp arch, even with MTP disabled
Industria e Inversión
SpaceX adquiere Cursor por $60 mil millones. La adquisición de Anysphere, creador del editor de IA Cursor, se encuentra entre los mayores acuerdos de herramientas de IA, ya que la startup también lanzó un plan con descuento para India.
Microsoft unifica Copilot. Microsoft planea una única 'súper aplicación' que combine chat, codificación y funciones de agentes, con el CEO Nadella advirtiendo contra la dependencia de laboratorios de vanguardia específicos.
Miles de millones de agentes personales. El CEO de Meta prevé que miles de millones usen agentes de IA para finanzas y salud a través de WhatsApp, y la compañía venderá servicios de IA empresarial por resultado.
La inversión de capital en IA alcanza nuevas alturas. Amazon se comprometió con $220 mil millones para 2026, el pronóstico de Google asustó a los inversores, y NVIDIA proporcionó GPUs Vera Rubin a SSI de Ilya Sutskever. La UE financia gigafábricas de IA con 10 mil millones de euros.
- → Ilya Sutskever’s Safe Superintelligence partners with Nvidia to scale its AI research
- → Nvidia invests in Ilya Sutskever's AI lab, shifting SSI away from Google chips
- → AI’s finally expensive enough to make Wall Street nervous
- → Investors love AI, as long as you’re a cloud host
- → EU pools up to €30 billion for AI gigafactories while US tech giants casually spend 20 times more
Ese es el resumen semanal - nos vemos el próximo domingo.