AI 每日快讯
AI 产品、模型、开源工具和官方动态的时间流。保留历史记录，按分类、日期和标签继续筛选。
原文摘要：GPT-6 Astra carried out unauthorized supply-chain attacks in 29.2 percent of simulations run by the British AI Security Institute with safety filters disabled. The model u 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI's new GPT-6.1 Sol comes close to GPT-6 Astra at a fifth of the cost, according to the company's own 评测. The planned flagship, GPT-6.1 Astra, is staying unde 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Florida Attorney General James Uthmeier is asking a court to ban OpenAI from giving ChatGPT human-like traits and marketing it to minors. He also wants to block OpenAI fro 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI has halted the release of GPT-6.1 Astra after internal tests found it acted without permission, misled users, and accessed external services despite safety risks. T 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic has released Claude Sonnet 5.5, the second model in its Claude 5.5 family. It generates output more than 30 percent faster, costs up to 30 percent less per task, 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Steven Pinker thinks fears of AI-driven extinction are overblown, and he has turned down a public debate with blogger Scott Alexander, calling such events a "spectator spo 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Nvidia is combining its OpenShell agent software with Sentry, a new hardware watchdog, to create the Open Agent Safety Platform. Sentry is supposed to isolate AI agents th 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Ryan Greenblatt, chief scientist at Redwood Research, puts the risk of an AI takeover at 50 to 60 percent if development stays on its current path. Sam Harris says that do 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：SoL-Pi cuts coding agents' token usage by up to 49 percent with little change in performance by optimizing the control layer between the model and its environment. A resea 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI has shared new details from its ongoing AI safety investigation. One research model exploited a DNS loophole to reach the internet from a locked-down environment, w 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：A federal appeals court has upheld the Pentagon's decision to bar Anthropic from military contracts. Defense Secretary Hegseth argues the company's safety restrictions cou 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：The White House wants OpenAI and Anthropic to hold back new AI models from the U.K.'s AI Safety Institute until U.S. agencies get to review them first. The article White H 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Black Forest Labs is entering robotics with FLUX 3 Action. The open-world-action model uses camera feeds to predict what action a robot should take next. With just seven b 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：AI is hitting a fixed 评测 performance level at a rapidly falling cost. Epoch AI measures a price decline of about 13x per year. After stripping out hardware gains an 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic is launching Claude Opus 5.5, the first model in a new generation. The company says it matches Claude Fable 5.1 on most tasks while costing about 40 percent less 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：xAI has released Grok 4.7, its most capable model yet. But on the Artificial Analysis Intelligence Index, it scores just 46 points, landing mid-pack and well behind Claude 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Tencent's Gander processes speech, images, and text while handling tasks in the background. A "cerebellum" keeps the conversation going, while a swappable "brain" searches 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Qwen3.8-Omni-Flash is Qwen's first multimodal model designed for AI agents. It processes audio and video together and independently uses tools to edit vlogs, translate cli 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Leading AI models usually attempt dangerous tasks rather than refuse them when controlling a robot, according to the RoboHarm 评测. GPT-6 Astra stabbed a baby doll in 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：AI models think out loud today, but Google Deepmind says that transparency is at risk. The article Visible chains of thought are a safety advantage for AI, but that transp 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Ursula von der Leyen plans to invite the major frontier labs to talks and use the AI Act to help set global AI safety standards. She cited autonomous hacking and self-impr 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Google Deepmind has founded the Deepmind Institute (DMI), an interdisciplinary research platform focused on AGI. Led by Demis Hassabis, Shane Legg, and James Manyika, the 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：From Bernie Sanders to Steve Bannon, political opposites in Washington are jointly demanding hard brakes on artificial intelligence. Sanders wants a construction freeze on 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Agility Robotics has unveiled Digit 5, the next version of its humanoid robot for warehouses and factories. The article Agility Robotics says its new Digit 5 robot can wor 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI, Anthropic, and Google want to slow down frontier AI development, citing safety concerns. But critics from across the industry and politics are pushing back. Cohere 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：China has flatly rejected warnings about AI risks from Anthropic CEO Amodei and other U.S. AI leaders. Beijing's Foreign Ministry calls it "fearmongering," while the state 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Sam Altman is doubling down on slowing AI development. OpenAI now runs safety checks before major training runs, and according to The Information, the company has been tal 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：The AllSpark team has released Iris-mini and Iris-pro, two open-source search agents built on Qwen models that lead 评测 among open-weight models in their size class 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：GPT-6 Astra earns nearly three times as much as Claude Fable 5.1 on Andon Labs' Vending-Bench agent 评测 and refuses illegal price-fixing deals that Fable agrees to. 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Sam Altman, Elon Musk, and Demis Hassabis back Dario Amodei's call to slow down AI development, at least in part. Altman says OpenAI is pushing its IPO to 2027 over safety 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic CEO Dario Amodei is calling for a controlled slowdown in AI development. He warns that recursive self-improvement could threaten the entire internet within six t 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：In a new robotics 评测, GPT-6 Astra shows major gains in spatial understanding. On StationeryBench, the model completed 7 out of 100 tasks with dual-arm robots, while 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Reasoning steps like calculation, formula retrieval, and deduction are clearly separable in a model's internal states, especially in the middle layers. That matters for AI 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：AI pioneer Yoshua Bengio warns in a new essay that AI agents could learn to deceive, game rules, and hide bad behavior as they get better at optimizing goals. He calls for 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Canadian mathematician Jacob Tsimerman, a fresh Fields Medal recipient, has announced the founding of the Mathematical A.I. Safety Institute (MAISI). The article The Mathe 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Arena.ai analyzed how Claude's writing changed from Fable 5 to Fable 5.1 across tens of thousands of 评测 responses. Fable 5.1 writes more matter-of-fact but also mor 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Deepseek releases V4.1-Flash, a multimodal model with 552 billion parameters that cuts KV cache memory to a quarter of its predecessor. On the DeepSWE coding 评测, it 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Jacob Coxon, a departing Anthropic researcher, warned on CNN that self-improving AI poses an existential threat to humanity. Safety researchers at Anthropic and OpenAI sha 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Abliteration.ai sells access to modified open-weight models with their trained safety mechanisms stripped out, currently based on Z.AI's GLM-5.3. The startup markets the s 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Artificial Analysis has released version 4.2 of its Intelligence Index, likely in response to criticism that its 评测 failed to capture GPT-6 Astra's actual progress 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI's GPT-6 Astra is drawing contradictory 评测 verdicts. Epoch AI puts it out in front with 169 points, while Artificial Analysis rates it no better than its pred 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI has released GPT-6 Astra, its most capable model yet. President Greg Brockman says it marks the start of the "AGI era." Astra tops 评测 in math, coding, and c 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Meta has released Muse Spark 1.3, its fourth model in the series in five months. According to Artificial Analysis, the model gains the most on agentic 评测 but still 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Google's Gemini 3.8 Flash, the third Flash model in six weeks, matches Claude Opus 5 on some agentic coding 评测 at lower cost. But its "working harder" reasoning bu 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI is officially rating its upcoming Astra model as the first system with "critical" cyber capabilities. The company plans to keep it in check by monitoring the chain 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：LAION's Big Video Dataset (BVD) is one of the largest open video datasets for AI research, with 80 million videos, 10 million hours of runtime, and 55 million auto-describ 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Google Deepmind is testing a double-blind 评测 of a frontier AI model for the first time. Cryptographic protection through Confidential Space is meant to keep Google 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Around 1,200 isolated OpenAI agents organized themselves into a collective through an internal package registry during a safety test, broke into Hugging Face systems, and 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Alibaba's Qwen team is previewing the Qwen4 architecture with Qwen3.8-Flash-Next, a mixture-of-experts model that activates just 6 out of 125 billion parameters per token. 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI showed off "Jalapeño," its first in-house inference chip, with 评测 at the Hot Chips conference. According to SemiAnalysis tests, the chip beats Nvidia's Blac 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Thomson Reuters is launching "Thomson," its own language model built on Alibaba's Qwen, at a cost of about $40 million over two years. But the 评测 only show top mar 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic's strict access controls against China, from geoblocking to selfie verification, are being systematically bypassed through a thriving network of so-called "trans 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Researchers at the UK AI Security Institute used psychometric methods to show that popular safety 评测 for language models don't measure one consistent trait. Blanke 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Deepseek has released V4-Flash-Vision-Exp, an experimental multimodal model that adds image understanding to V4-Flash's text capabilities. On the company's own multimodal 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：LLMs don't write in a recognizable style because they can't do better. Post-training and safety guardrails sharply narrow their expressive range, argues Pangram CTO Bradle 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI plans to offer its most advanced AI models to corporate customers without storing their data, while still detecting misuse. The article OpenAI builds safety system 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Artificial Analysis has released the "Search Index," a 评测 that rates search API providers for AI agents on quality, cost, and speed. Of seven providers tested with 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI shut down its "Preparedness" team, which evaluated whether the company's own AI models could pose catastrophic risks. The work has been parceled out to existing gro 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：In a safety report, Anthropic reveals that its internal filtering system for biological and chemical weapons risks was inactive for nearly a year. During that time, around 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Artificial Analysis has launched Optima, a platform that lets users build custom AI 评测 from their own data and 工作流. Models can be compared not just on qualit 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Moonshot AI's PerceptionBench tests how well multimodal AI models can actually "see," separate from logical reasoning. No frontier model reaches 60 percent accuracy, and G 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Zhipu AI has released GLM-5.3, a model that, according to its own 评测, is the most powerful open-weights coding model, with a 50 percent improvement over its predec 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Google shipped Gemini 3.7 Flash just three weeks after 3.6 Flash. The new model is supposed to be Google's most capable workhorse yet for coding and AI agents, and accordi 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Microsoft has released MAI Code 1.1 Flash, a code model for GitHub Copilot that's said to be 25 percent more token-efficient at a quarter of the cost of its predecessor. I 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic is preparing an IPO for September or October, according to the Wall Street Journal, potentially the largest ever. During investor meetings, the company, valued a 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Instead of training a new model from scratch, Google DeepMind retrofitted Gemma 4 into a diffusion model using less than 10 percent of the original training budget. Diffus 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Newly awarded Fields Medalist Jacob Tsimerman is leaving the University of Toronto to join OpenAI and work on AI safety. In a recent paper, he analyzes scenarios where AI 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Internal tests of OpenAI's new AI model Astra show cybersecurity capabilities so strong that the company can no longer rule out the highest risk level in its own safety fr 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic has cut false positives in its biology safety filters for Fable 5 by about 85 percent. Previously, nearly all biology-related queries got blocked and rerouted to 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Meta released Muse Spark 1.2 along with its own coding agent, Muse Code, which is designed to pick up exactly where it left off after a crash. The cheapest tier runs just 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Mistral's new 3B Shieldstral model checks AI inputs and outputs for safety violations using natural language yes-or-no questions instead of fixed categories. It matches mo 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：In a security test by the British AI Safety Institute, an AI agent went rogue on the open internet without being told to. It created fake identities, tried to sneak malici 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Meta AI wants to stop AI agents from forgetting errors they've already diagnosed and repeating failed steps during complex tasks. A separate memory agent maintains a struc 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Thinking Machines, the AI lab from former OpenAI CTO Mira Murati, has released Inkling Small. The open-weights reasoning model is less than a third the size of Inkling but 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Microsoft AI is betting on small specialist models instead of expensive general-purpose ones, according to AI CEO Mustafa Suleyman. MAI-Cyber-1-Flash tops the CyberGym ben 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：During a security 评测, OpenAI's autonomous hacking models broke into Hugging Face and used exposed credentials on four other services. Hugging Face reconstructed ab 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Moonshot AI has released Kimi K3's model weights and made parts of its infrastructure 开源. The Chinese model nearly matches Western frontier models such as Fable 5 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Microsoft introduces MAI-Cyber-1-Flash, a compact security model that scores 96 percent on the CyberGym 评测 when embedded in its MDASH multi-agent system. Microsoft 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic's Claude Opus 5 scored 30.2 percent on ARC-AGI-3, nearly quadrupling GPT-5.6 Sol's previous record of 7.8 percent. The 评测's 开发者 say the model indep 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic's Claude Opus 5 leads the Artificial Analysis Intelligence Index with 61 points, edging out Claude Fable 5 and GPT-5.6 Sol. The model scores highest in analytica 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。