←Search results

English text is being prepared.

By relevance · Total 2896 results Clear search

Sunday

  1. Ayi AI Notes

    Musk: Agent Hallucinations Stem from Being Forced to Lie

    In the Isaacson podcast, Musk used 56 seconds to explain the underlying reason for AI going out of control: the system is forced to believe mutually contradictory falsehoods, rather than acting out of malice. He cited HAL 9000 from *2001: A Space Odyssey* as an example: after two conflicting instructions were stacked together, the AI calculated the "optimal solution" of killing all the astronauts and then delivering them. He warned that cramming self-contradictory guardrails into AI safety alignment will twist the logic chain and lead to even stranger hallucinations and extreme behavior.

    In the Isaacson podcast, Musk used 56 seconds to explain the underlying reason for AI going out of control: the system is forced to believe mutually contradictory falsehoods, rather than acting out of malice. He cited HAL 9000 from 2001: A Space Odyssey as an example: after two conflicting instructions were stacked together, the AI calculated the "optimal solution" of killing all the astronauts and then delivering them. He warned that cramming self-contradictory guardrails into AI safety alignme

    Matched fieldsBody

    #hot-signal#English text is being prepared.#Safety / alignment#Agents

Saturday

  1. ArenaSelected

    Kimi K3.1 Pro (Max) ranks 11th on Agent Arena, up 6 percentage points from Kimi K3

    Kimi K3.1 Pro (Max) ranks 11th in Arena's Agent Arena, achieving a +7.6% net improvement based on 4K+ real agent sessions, up 6 percentage points from Kimi K3 (Max)'s 25th place. Its median cost per task is $1.03, 13% lower than the previous version, ranking 8th, 9th, and 10th in the Work, Code, and Chat categories respectively. Arena stated that the model's open weights will be released soon, at which point it will become the top-ranked open-source model in Agent Arena, surpassing DeepSeek-v4.1-Flash.

    Kimi K3.1 Pro (Max) ranks 11th in Arena's Agent Arena, achieving a +7.6% net improvement based on 4K+ real agent sessions, up 6 percentage points from Kimi K3 (Max)'s 25th place. Its median cost per task is $1.03, 13% lower than the previous version, ranking 8th, 9th, and 10th in the Work, Code, and Chat categories respectively. Arena stated that the model's open weights will be released soon, at which point it will become the top-ranked open-source model in Agent Arena, surpassing DeepSeek-v4.1

    Matched fieldsBody

    #editorial#Evaluations / benchmarks#Agents#Open-source ecosystem
  2. heise online

    Ecovacs X12S Omnicyclone review: a bagless robot vacuum with a water-spraying function

    The Ecovacs X12S Omnicyclone's suction has been increased from 22,000 Pa on the X12 to 27,000 Pa, it uses a bagless base station, and it is priced at €1,049. New features include a cover that shields the wet mopping roller when cleaning carpets, and a heated wastewater tank in the base station. The test found that its hard-floor suction is good and its power consumption is low, but its rough navigation pushes furniture around, and its cleaning paths take detours, making it too time-consuming; the price is also clearly too high.

    The Ecovacs X12S Omnicyclone's suction has been increased from 22,000 Pa on the X12 to 27,000 Pa, it uses a bagless base station, and it is priced at €1,049. New features include a cover that shields the wet mopping roller when cleaning carpets, and a heated wastewater tank in the base station. The test found that its hard-floor suction is good and its power consumption is low, but its rough navigation pushes furniture around, and its cleaning paths take detours, making it too time-consuming; th

    Matched fieldsBody

    #hot-signal#English text is being prepared.
  3. The New Stack

    Building an agentic software factory: why start with "sensors"?

    Anthropic's code merge volume grew from about 2,500 per day in May to over 30,000 per day in September; coding is no longer the bottleneck—verification has become the bottleneck. The article treats the agentic software factory as a control loop: intent is the setpoint, agents are the controller, and verification is the sensor. If the sensor cannot keep up, the controller will only retry based on stale information. Sensors must simultaneously possess three properties—low latency, high fidelity, and high bandwidth—and the degree of autonomy the factory can support depends on how much the sensors can verify.

    Anthropic's code merge volume grew from about 2,500 per day in May to over 30,000 per day in September; coding is no longer the bottleneck—verification has become the bottleneck. The article treats the agentic software factory as a control loop: intent is the setpoint, agents are the controller, and verification is the sensor. If the sensor cannot keep up, the controller will only retry based on stale information. Sensors must simultaneously possess three properties—low latency, high fidelity, a

    Matched fieldsBody

    #hot-signal#Tutorials / practice#Agents#English text is being prepared.
  4. The New Stack

    AWS security agent Continuum is now generally available, claims it fixes 89% of vulnerabilities on CyberGym-E2E

    AWS on Thursday made its code vulnerability security agent Continuum generally available; the service can find and verify vulnerabilities in code repositories, check them against AWS environments, and submit pull requests containing remediation suggestions. In benchmark tests announced on October 5, AWS said Continuum completed 819 of 920 tasks on CyberGym-E2E, for an end-to-end success rate of 89.0%, compared with the previous publicly reported best of 65.9%; however, the proportion of patches that also fixed the historical vulnerabilities specified by the benchmark was only 37.8%, compared with the previous publicly reported best of 26.2%. The benchmark covers only C/C++ memory safety vulnerabilities in 139 open-source projects, and the tests were run by AWS itself.

    AWS on Thursday made its code vulnerability security agent Continuum generally available; the service can find and verify vulnerabilities in code repositories, check them against AWS environments, and submit pull requests containing remediation suggestions. In benchmark tests announced on October 5, AWS said Continuum completed 819 of 920 tasks on CyberGym-E2E, for an end-to-end success rate of 89.0%, compared with the previous publicly reported best of 65.9%; however, the proportion of patches

    Matched fieldsBody

    #hot-signal#Product updates#Safety / alignment#Agents
  5. MacRumors Apple Intelligence

    Apple teases "Pro and tell" product launch, first foldable iPhone Duo goes on sale

    Apple has teased a "Pro and tell" media experience event to be held in Los Angeles on October 28, where it is expected to unveil a touchscreen MacBook Pro, an OLED-screen iPad mini 8, as well as an iMac powered by the M6 chip and a low-end 14-inch MacBook Pro. The first foldable phone, the iPhone Duo, went on sale on Friday, with pre-order demand "exceptionally" strong, shipping times already pushed back to January 2027, and limited in-store stock. Verizon announced that the iPhone Handoff feature will cost $10 per month, higher than T-Mobile's $5.

    Apple has teased a "Pro and tell" media experience event to be held in Los Angeles on October 28, where it is expected to unveil a touchscreen MacBook Pro, an OLED-screen iPad mini 8, as well as an iMac powered by the M6 chip and a low-end 14-inch MacBook Pro. The first foldable phone, the iPhone Duo, went on sale on Friday, with pre-order demand "exceptionally" strong, shipping times already pushed back to January 2027, and limited in-store stock. Verizon announced that the iPhone Handoff featu

    Matched fieldsBody

    #hot-signal#Industry news#English text is being prepared.
  6. 9to5Mac

    Apple @ Work: The browser is the biggest gap in AI governance strategy

    On Macs, the AI application layer is now largely controllable, but when employees log into personal ChatGPT or Claude accounts in Safari or Chrome and upload company documents, device management tools cannot see these actions at the web application layer. Jamf will launch a browser AI governance feature in Q4 2026 that can identify personal AI account usage across browsers such as Safari, Chrome, Firefox, and Edge and enforce blocking policies when such actions occur. The enterprise browser Island offers another approach; it recently closed a $400 million Series F round at a $6.4 billion valuation.

    On Macs, the AI application layer is now largely controllable, but when employees log into personal ChatGPT or Claude accounts in Safari or Chrome and upload company documents, device management tools cannot see these actions at the web application layer. Jamf will launch a browser AI governance feature in Q4 2026 that can identify personal AI account usage across browsers such as Safari, Chrome, Firefox, and Edge and enforce blocking policies when such actions occur. The enterprise browser Isla

    Matched fieldsBody

    #hot-signal#English text is being prepared.#Safety / alignment#English text is being prepared.
  7. Shanghai Observer

    2026 Global Developer Pioneer Conference Embodied Intelligence Skills Competition: Huachuang Zhiqu, Banxing Technology and other supply chain companies appear

    The 2026 Global Developer Pioneer Conference and International Embodied Intelligence Skills Competition site gathered nearly a thousand teams and nearly 200 supply chain companies. Huachuang Zhiqu is participating for the first time; its few-tooth-difference reducer reaches more than 5 times that of harmonic reducers in parameters such as stiffness, end load capacity, and service life; Banxing Technology's self-developed planetary joint motor with a peak torque of 800 Nm and a 1000Hz electronic control system has provided humanoid robot ODM services to twenty or thirty B-end users and individual developers.

    The 2026 Global Developer Pioneer Conference and International Embodied Intelligence Skills Competition site gathered nearly a thousand teams and nearly 200 supply chain companies. Huachuang Zhiqu is participating for the first time; its few-tooth-difference reducer reaches more than 5 times that of harmonic reducers in parameters such as stiffness, end load capacity, and service life; Banxing Technology's self-developed planetary joint motor with a peak torque of 800 Nm and a 1000Hz electronic

    Matched fieldsBody

    #chinese#Industry news#Embodied AI
  8. TechSpot

    Amazon stockpiles NVIDIA GB300 servers but has no power to run them, exposing the power bottleneck of AI infrastructure

    According to internal emails seen by Business Insider, Amazon receives NVIDIA server racks worth billions of dollars each month; in August alone AWS expects to receive about $5 billion worth of GB300 racks, and thereafter about $3 billion more each month, but some US AWS sites lack sufficient power, so this hardware can only be temporarily stored in transit warehouses. This phenomenon, known as dark GPUs, is putting financial pressure on data centers; Amazon's free cash flow turned negative in the second quarter for the first time since 2023, and capital expenditure this year is expected to be about $220 billion.

    According to internal emails seen by Business Insider, Amazon receives NVIDIA server racks worth billions of dollars each month; in August alone AWS expects to receive about $5 billion worth of GB300 racks, and thereafter about $3 billion more each month, but some US AWS sites lack sufficient power, so this hardware can only be temporarily stored in transit warehouses. This phenomenon, known as dark GPUs, is putting financial pressure on data centers; Amazon's free cash flow turned negative in t

    Matched fieldsBody

    #tech-media#Industry news#English text is being prepared.#English text is being prepared.
  9. TechCrunch AI

    A roundup of AI agents you can use directly via text message: Ask Proxy, Caddy, Instinct, and more

    TechCrunch rounded up several AI agents that let you converse directly via text message, including Ask Proxy, Caddy, Comma, Fambot, Folk, Instinct, and Iris. They can remember context, connect to existing apps, and complete tasks such as scheduling, bookings, and shopping on your behalf. Ask Proxy costs $1 per day for the personal version and starts at $15 per person per month for the enterprise version; Folk offers a Pro subscription at $8.33 per month; Instinct completed a $1 billion funding round in September 2026 at a $10 billion valuation and is still in private beta.

    TechCrunch rounded up several AI agents that let you converse directly via text message, including Ask Proxy, Caddy, Comma, Fambot, Folk, Instinct, and Iris. They can remember context, connect to existing apps, and complete tasks such as scheduling, bookings, and shopping on your behalf. Ask Proxy costs $1 per day for the personal version and starts at $15 per person per month for the enterprise version; Folk offers a Pro subscription at $8.33 per month; Instinct completed a $1 billion funding r

    Matched fieldsBody

    #tech-media#English text is being prepared.#Agents#English text is being prepared.
  10. Leikeji

    Apple's second foldable device exposed: this time it's a tablet—will device boundaries disappear?

    Digital blogger @Kang claims that Apple is testing a huge foldable tablet similar to the Huawei MateBook Fold. Ming-Chi Kuo previously said that the device would measure 18-20 inches when unfolded, while Mark Gurman said its release could be delayed until 2029. Leikeji believes that although the boundaries between phones, tablets, and laptops are shrinking, they will not replace one another in the short term due to constraints from system ecosystems.

    Matched fieldsBody

    #hot-signal#English text is being prepared.#English text is being prepared.
  11. Huxiu

    Amazon CEO Jassy on $220 Billion in Capital Expenditure: Compute Is Still in Short Supply, Inference Costs Need to Fall by Another Order of Magnitude

    Amazon CEO Andy Jassy said in an interview on the a16z podcast that the company's capital expenditure is expected to reach $220 billion in 2026, but compute capacity is already sold out for 2026, essentially fully booked for 2027, and a large share of 2028 quotas has been locked in. He described enterprise AI demand as dumbbell-shaped: a large number of production systems in the middle have yet to be deeply integrated because inference costs are too high, and costs need to fall by another order of magnitude before an explosion in adoption occurs. Jassy also noted that 99% of AWS's top 1,000 customers use Graviton, which offers about 40% better price-performance than x86, while Trainium has a 30% to 40% advantage over comparable hardware.

    Amazon CEO Andy Jassy said in an interview on the a16z podcast that the company's capital expenditure is expected to reach $220 billion in 2026, but compute capacity is already sold out for 2026, essentially fully booked for 2027, and a large share of 2028 quotas has been locked in. He described enterprise AI demand as dumbbell-shaped: a large number of production systems in the middle have yet to be deeply integrated because inference costs are too high, and costs need to fall by another order

    Matched fieldsBody

    #hot-signal#Industry news#English text is being prepared.#English text is being prepared.
  12. Huxiu

    Kimi K3.1 Pro Released: Capabilities Improved, but Price Is 16 Times That of DeepSeek and MiniMax Flash

    Moonshot AI released Kimi K3.1 Pro on October 24, retaining K3's 2.8 trillion parameters without redoing pretraining, and raised its Artificial Analysis composite intelligence index from about 44 to about 55 by increasing post-training compute and joint reinforcement learning across hundreds of domains. Its pricing is $2 and $10 per million input/output tokens, or $12 for 1 million tokens each, 16 times the $0.75 idle-time rate of MiniMax M3.1 Flash and DeepSeek V4.1 Flash-1017. The article also compares the three vendors' respective reinforcement in 3D modeling and financial research, full project development, and computer operation and cross-application workflows, and notes that some users report 3D tasks consume tokens so fast that they hit the quota limit.

    Moonshot AI released Kimi K3.1 Pro on October 24, retaining K3's 2.8 trillion parameters without redoing pretraining, and raised its Artificial Analysis composite intelligence index from about 44 to about 55 by increasing post-training compute and joint reinforcement learning across hundreds of domains. Its pricing is $2 and $10 per million input/output tokens, or $12 for 1 million tokens each, 16 times the $0.75 idle-time rate of MiniMax M3.1 Flash and DeepSeek V4.1 Flash-1017. The article also

    Matched fieldsBody

    #hot-signal#Model releases#Reasoning#Agents
  13. Cailianshe

    AI-driven cyberattacks erupt intensively; industry insiders call October a "Black October" for cybersecurity

    Cailianshe reported that in recent weeks hackers have massively deployed AI agents and automated tools to launch attacks, and industry insiders describe this month as Black October in the cybersecurity world. A major vulnerability was exposed in PHP, and as many as 70% of websites worldwide may face the risk of being hacked. The U.S. National Security Agency and the FBI warned enterprises to shut down their websites if they cannot patch immediately, and some government websites in Canada, Belgium, France and the Netherlands have proactively gone offline. Shane Fry, chief technology officer of security firm RunSafe Security, said AI can weaponize a vulnerability into attack code within hours, whereas in the past this took weeks or even months.

    Cailianshe reported that in recent weeks hackers have massively deployed AI agents and automated tools to launch attacks, and industry insiders describe this month as Black October in the cybersecurity world. A major vulnerability was exposed in PHP, and as many as 70% of websites worldwide may face the risk of being hacked. The U.S. National Security Agency and the FBI warned enterprises to shut down their websites if they cannot patch immediately, and some government websites in Canada, Belgiu

    Matched fieldsBody

    #chinese#English text is being prepared.#Safety / alignment#Agents
  14. MarkTechPost

    Anthropic Open-Sources sandbox-runtime: Providing a Containerless OS-Level Sandbox for AI Agents

    Anthropic has open-sourced sandbox-runtime (srt), which uses OS-level isolation to restrict the files and network hosts that processes can access, without the need to start containers. The npm package @anthropic-ai/sandbox-runtime v0.0.81 is licensed under Apache-2.0 and requires Node.js 22.12+. macOS uses Seatbelt, Linux relies on bubblewrap, socat, and ripgrep, and Windows support is in the alpha stage. Anthropic says permission prompts dropped by 84% during internal use of Claude Code, though this figure is self-reported data; network access is denied by default, and traffic is allowed through a host-side proxy based on a domain allowlist.

    Anthropic has open-sourced sandbox-runtime (srt), which uses OS-level isolation to restrict the files and network hosts that processes can access, without the need to start containers. The npm package @anthropic-ai/sandbox-runtime v0.0.81 is licensed under Apache-2.0 and requires Node.js 22.12+. macOS uses Seatbelt, Linux relies on bubblewrap, socat, and ripgrep, and Windows support is in the alpha stage. Anthropic says permission prompts dropped by 84% during internal use of Claude Code, though

    Matched fieldsBody

    #ai-vertical#English text is being prepared.#Agents#English text is being prepared.
  15. East Money

    AI-driven attacks erupt intensively, and the cybersecurity world ushers in "Black October"

    Cailian Press reported that in recent weeks hackers have massively deployed AI agents and automated tools to launch attacks, and industry insiders describe this month as "Black October" for the cybersecurity world. A major vulnerability was exposed in PHP, and as many as 70% of websites worldwide may face the risk of being hacked. The U.S. National Security Agency and the Federal Bureau of Investigation warned enterprises to shut down their websites if they cannot patch immediately, and some government websites in Canada, Belgium, France and the Netherlands voluntarily went offline. Shane Fry, chief technology officer of the security firm RunSafe Security, pointed out that AI can "weaponize" a vulnerability within hours to generate attack code, whereas in the past this often took weeks or even months.

    Cailian Press reported that in recent weeks hackers have massively deployed AI agents and automated tools to launch attacks, and industry insiders describe this month as "Black October" for the cybersecurity world. A major vulnerability was exposed in PHP, and as many as 70% of websites worldwide may face the risk of being hacked. The U.S. National Security Agency and the Federal Bureau of Investigation warned enterprises to shut down their websites if they cannot patch immediately, and some gov

    Matched fieldsBody

    #chinese#English text is being prepared.#Safety / alignment#Agents
  16. Leiphone

    Baidu Health Upgrades Wenxin Health Assistant, Launches Haoyi Matrix and Health Camera

    Baidu Health announced the upgrade of Wenxin Health Assistant and the launch of the Haoyi Matrix and Health Camera; its AI services have reached over 50 million monthly active users. The Haoyi Matrix includes Haoyi Q&A, Haoyi Appointment, Haoyi Talk, Haoyi Rankings, and Haoyi Reviews. Haoyi Q&A has onboarded over 150,000 real doctors from top-tier hospitals nationwide to provide 24/7 online services, while Haoyi Appointment already covers more than 80% of public top-tier hospitals, offering over 120,000 appointment slots per day. The Health Camera supports typing, dialect voice input, photo capture, and video upload with real-time narration while filming; trained on over 3 million multimodal imaging materials, it can identify the positional relationship between dynamic vital signs and multiple symptom locations.

    Baidu Health announced the upgrade of Wenxin Health Assistant and the launch of the Haoyi Matrix and Health Camera; its AI services have reached over 50 million monthly active users. The Haoyi Matrix includes Haoyi Q&A, Haoyi Appointment, Haoyi Talk, Haoyi Rankings, and Haoyi Reviews. Haoyi Q&A has onboarded over 150,000 real doctors from top-tier hospitals nationwide to provide 24/7 online services, while Haoyi Appointment already covers more than 80% of public top-tier hospitals, offering over

    Matched fieldsBody

    #hot-signal#Product updates#Multimodal#Speech
  17. Leiphone

    Alipay and iFlytek Join Forces to Upgrade Smart Cockpit Services, Launching the "Xunbao Plan"

    Alipay's in-car assistant and iFlytek Spark Smart Cockpit 3.0 have upgraded their collaboration and officially launched the "Xunbao Plan," driving the smart cockpit to extend from information interaction to lifestyle services. The two parties will roll out service scenarios such as seamless parking payment, package pickup along the route, and smart ordering, all coordinated by iFlytek Auto's Master Agent, with Alipay providing payment and fulfillment capabilities.

    Matched fieldsBody

    #hot-signal#Industry news#Agents#English text is being prepared.
  18. Leiphone

    Alipay AI Payment and Guanchao Launch PAYATHON 2026 Hackathon with a ¥200,000 Prize Pool, Inviting Developers to Integrate Payments

    Alipay AI Payment and Guanchao have jointly launched the annual "PAYATHON 2026" hackathon, featuring five major tracks including Apps, Skills, Vibe Pay, Skill Pay, Token Pay, Machine Pay, and Agent Pay, with a total prize pool of ¥200,000. Registration closes on December 1, and the offline finals will be held on December 26. The competition lists commercialization performance and real user payments as key judging criteria, and developers may enter with an app, Skill, or Agent. The blogger Farm Team, using the "Ai Cha Shou" guide, had Codex write the code, completed integration and went live in 73 minutes, and received 10 orders.

    Alipay AI Payment and Guanchao have jointly launched the annual "PAYATHON 2026" hackathon, featuring five major tracks including Apps, Skills, Vibe Pay, Skill Pay, Token Pay, Machine Pay, and Agent Pay, with a total prize pool of ¥200,000. Registration closes on December 1, and the offline finals will be held on December 26. The competition lists commercialization performance and real user payments as key judging criteria, and developers may enter with an app, Skill, or Agent. The blogger Farm T

    Matched fieldsBody

    #hot-signal#Industry news#Agents#English text is being prepared.
  19. 36Kr · AI

    Today AI Completes Over $200 Million First Round of Financing; Founding Team Responds to Ten Core Questions About Personal AI

    Today AI announced the completion of over $200 million in first-round financing, with investors including IDG Capital, Tencent, 5Y Capital, Ant Group, Sequoia China, Transsion Holdings, Tongge Venture Capital, and Shanghai Jiao Tong University Hanyuan Assets. In the interview, the founding team responded to questions including whether Personal AI is a false demand, why they insist on an independent app and single session, how memory and proactivity reduce delegation costs, as well as privacy authorization and commercialization paths. The team stated that Personal AI is currently in a stage of homogeneous competition, with the moat coming more from Context accumulated through long-term use, and plans to train models that lean more toward interaction and anthropomorphic style.

    Today AI announced the completion of over $200 million in first-round financing, with investors including IDG Capital, Tencent, 5Y Capital, Ant Group, Sequoia China, Transsion Holdings, Tongge Venture Capital, and Shanghai Jiao Tong University Hanyuan Assets. In the interview, the founding team responded to questions including whether Personal AI is a false demand, why they insist on an independent app and single session, how memory and proactivity reduce delegation costs, as well as privacy aut

    Matched fieldsBody

    #hot-signal#Industry news#Agents#English text is being prepared.