독서리서치레이더투자 프레임워크
로그인 / 회원가입
전체기술제품투자상장시장기타

독서

2026-11-1627 게시물

새로고침

Security firm Island reports over 180 malicious domains impersonating ChatGPT and other AI brands

주제 · Island중대
2026-11-16 23:18 GMT+8

Security research firm Island has reported the discovery of 186 "typosquatting" domains designed to impersonate 27 popular AI brands, including ChatGPT, Claude, Perplexity, Midjourney, and Cursor. These sites exploit common user misspellings in address bars (such as typing chtgpt.com instead of chatgpt.com) to redirect victims to malicious destinations.

According to Island's findings, these domains serve three primary attack vectors depending on the victim's location, operating system, and browser: ClickFix attacks disguised as Cloudflare verification pages that trick Windows users into executing malicious PowerShell commands; a seemingly legitimate Microsoft Store-signed app called "Net Guard" that actually hijacks browsers, modifies default search engines, and executes remote JavaScript; and prompts to download malicious browser extensions used for search traffic monetization or data theft.

While some domains have been seized, the cleanup is incomplete. As of November 3, 2026, 17 of the 30 identified malicious browser extensions remained listed on the Chrome Web Store. Island warns that the campaign remains active, advising users to bookmark their favorite AI tools and carefully verify URLs before visiting.

출처:TechRadar

리서치

Tesla Discloses Restrictive Terms for Personal Cybercab Sales: Mandatory Subscription, Geofencing, and Remote Brick Risk

중대
2026-11-16 21:07 GMT+8

U.S. electric vehicle maker Tesla has disclosed the terms and conditions for the invite-only personal sale of its autonomous robotaxi, the Cybercab. According to order and subscription agreements reviewed by Business Insider, the Cybercab starts at $29,990 but requires an additional "autonomy subscription" of $299 per month. Without this subscription, the vehicle will not operate, effectively rendering it a useless ornament.

These terms highlight Tesla's strategic shift from selling cars as one-time assets to providing them as continuous service gateways. Beyond the high fixed monthly cost, usage is strictly limited. Owners can only drive within the operational areas of Tesla's Robotaxi service, which currently includes parts of Austin and Dallas in Texas, and several cities in Florida. Notably, despite Elon Musk's previous pitches about owners earning money by adding their idle vehicles to the fleet, the new terms explicitly prohibit individual owners from generating revenue through rental services.

Tesla has also erected significant barriers regarding ownership disposal. Owners are barred from reselling the Cybercab without company permission, facing a $50,000 penalty if they do. More unusually, the agreement states that Tesla may remotely "brick" the vehicle, disabling it entirely, if violations occur. Additionally, Tesla retains the right to repurchase the vehicle at any time at a fair market value determined by the company, with no guarantee of a replacement, while owners bear the shipping costs back to Tesla.

Data privacy is another component of the terms. The subscription agreement indicates that Tesla will collect personally identifying data, including location and cabin camera image data, to provide ride support and improve the product. It remains unclear whether Cybercab owners can opt out of this data collection mechanism.

This series of restrictive clauses marks a further tech-company pivot in Tesla's business model. As traditional models like the Model S, X, and the soon-to-be-discontinued Cybertruck fade, Tesla is attempting to reshape automotive profitability through software subscriptions and service control. For early invited users, this means purchasing a controlled service terminal rather than a fully discretionary mode of transport.

출처:Business Insider

리서치

SAP Completes Acquisition of TechWolf to Integrate Context Graph into SuccessFactors for Agent Optimization

중대
2026-11-16 23:00 GMT+8

German software giant SAP SE (NYSE: SAP) announced on November 16, 2026, that it has completed the acquisition of TechWolf, a provider of AI work intelligence platforms. This transaction marks a substantive step in SAP's effort to deeply integrate artificial intelligence into core enterprise business processes, particularly in human resources and workforce planning.

According to SAP's announcement, TechWolf is expected to become an "intelligent core" of the SAP SuccessFactors portfolio, supporting the company's talent and work intelligence strategy. The key technical asset from this acquisition is TechWolf's proprietary "context graph." SAP plans to use this graph as the grounding layer for its AI assistant, Joule, when handling queries related to work and skills planning and talent management.

For enterprise customers, the direct expected benefit of this architectural change lies in efficiency and cost. SAP stated that more precise contextual understanding will make token usage more efficient, thereby lowering the cost of deploying workforce AI agents and enhancing Joule's performance in AI-driven HR scenarios. SAP had previously disclosed the intention to acquire TechWolf on October 6, 2026, stating that the move would provide enterprises with an evidence-based view of work.

Although SAP did not disclose specific financial terms, this acquisition reflects a trend among major ERP vendors shifting from merely providing model interfaces to building vertical-specific data infrastructure. By internalizing TechWolf's technology, SAP aims to address potential hallucinations or context gaps that general-purpose large models might exhibit in specific enterprise workflows, offering its vast customer base a more controllable and cost-effective AI application experience.

출처:SAP News Center

리서치

Ex-Freshworks Execs Raise $53M for Dreamteam's AI-Native CRM to Challenge Salesforce and HubSpot

주제 · Dreamteam중대
2026-11-16 20:30 GMT+8

Dreamteam, a San Francisco-based startup founded by former Freshworks executives, has raised $53 million across seed and Series A rounds, led by Together Fund and Accel respectively, with participation from Premji Invest. The company declined to disclose the size of each round or its valuation. The funding supports its push into the customer relationship management (CRM) market with an "AI-native" platform designed to eliminate the manual data entry burdens associated with traditional CRMs.

At the core of Dreamteam’s product is the "Company Brain," a virtual memory system that automatically captures context from emails, meetings, and other interactions to build a shared record of customer relationships. Four built-in AI agents use this information to research prospects, draft follow-ups, flag deal risks, and configure workflows, while sales teams retain control over outbound communications. Co-founder and CEO Anand Venkatraman stated that after more than 25 years in enterprise software, he had never seen a CRM provide a reliable forecast with a single click, prompting him to co-found the company in 2025 with Senthil Kanthaswamy and Ramesh Parthasarathy.

In terms of traction, Venkatraman reported that Dreamteam has signed up more than 100 customers across 13 countries in the three months since it began actively selling, with most paying for the service. Several businesses, including healthcare company DMS Health and media streaming firm Plex, have replaced established platforms like Salesforce and HubSpot. Scott Olechowski, Chief Product Officer at Plex, cited Dreamteam’s integration with existing AI tools as a "game changer." However, venture capital firm Andreessen Horowitz (a16z) argued earlier this year that while AI could transform how businesses manage customer relationships, existing CRMs are likely to remain the systems where records are stored, with new AI tools capturing value by working on top of them.

Dreamteam plans to expand beyond sales into other parts of the customer lifecycle, including onboarding, support, and upselling. The startup currently employs more than 35 people, with engineering based in Chennai and go-to-market operations in the United States. Kanthaswamy believes startups have a two-to-five-year window to gain ground before incumbent CRM providers can fully adapt their architectures to AI.

출처:TechCrunch

리서치

Google Maps Launches AI Trip-Planning Feature in Ask Maps

중대
2026-11-16 22:00 GMT+8

Google has announced a new AI trip-planning feature within its Maps application. Integrated into the Gemini-powered Ask Maps chatbot, this feature allows users to create custom travel itineraries and provides real-time details about each stop.

Users can plan one-day outings or multi-day trips using Ask Maps. For example, they can request a day trip for holiday shopping or a three-day road trip schedule with stops for coastal hikes and oceanfront dining. Ask Maps will generate a day-by-day schedule with a list of stops, showing the route and numbered pins for each location on the map. Users can share these itineraries with friends or family and export them to Google Docs on desktop.

In addition to trip planning, Ask Maps can now help users find specific appointments for beauty, pet care, auto, and home repair services, and call the locations on their behalf. Users can also use Ask Maps to search saved place lists, asking questions like “Which spots on my saved lists have availability for five people in the next 30 minutes?”

The update is rolling out to over 30 countries in Europe, including Spain, Germany, and the UK. Trip-planning and saved list search are being introduced in more than 200 countries where Ask Maps is available, while “know before you go” tips and appointment-finding are launching first in the US.

출처:The Verge

리서치

Audible Launches AI Audio Discovery Feature Combining Samples and Commentary

주제 · Audible간단한 요점
2026-11-16 22:00 GMT+8

On November 16, 2026, Audible announced the launch of a new "audio discovery" feature designed to help users determine whether they will enjoy an audiobook before starting to listen. The feature combines audiobook samples, AI-generated overviews, and commentary from authors or Audible editors to provide richer decision-making information than traditional text descriptions.

According to The Verge, the feature plays a snippet of the audiobook followed by a brief AI-generated overview that includes themes and listener reactions drawn from publicly available data like reviews. It also pulls direct quotes from past Audible interviews with authors or editors. Andy Tsao, Audible's Chief Product Officer, stated that picking your next listen should feel like getting a tip from a friend, and the goal is to make these quick audio takes more accessible.

In addition to audio discovery, Audible is using AI to personalize audiobook collections based on individual listening tastes and introducing AI-curated recommendation categories. These features are currently rolling out to select English-language titles in the US, with plans to expand to other regions. Audible is also expanding its beta features "Ask a Question," which allows users to query plot points and characters, and "Recaps," which provides spoiler-free audio summaries.

출처:The Verge

리서치

Siemens Announces Integration of Industrial AI and Additive Manufacturing for Prompt-Based Engineering

중대
2026-11-16 18:24 GMT+8

German industrial giant Siemens announced on November 16, 2026, that it is combining its "Industrial AI" capabilities with additive manufacturing to create an end-to-end Digital Thread. This initiative marks the first time agentic, prompt-based engineering is connected with an adaptive production environment, allowing individual product requirements to flow seamlessly from design through to manufacturing.

As a demonstration case, Siemens collaborated with robotics company AGIBOT to show how an Agentic AI workflow can customize robotic grippers for humanoid robots based on specific tasks. Users simply describe the object to be handled and the desired task. Through MCP integration, the AI agent accesses functions within engineering software like Siemens' Designcenter to select a suitable gripper type, adapt the geometry, simulate it, validate the robot interface, and finally prepare the data for additive manufacturing. This process shifts part of the engineering workflow from manual step-by-step adaptation to prompt-based orchestration.

Beyond design, Siemens plans to extend this adaptability to production. The company is working with additive manufacturing equipment provider EOS and post-processing technology specialist AM Solutions to create an "adaptive production cell" at its Electronics Factory in Erlangen, Germany. This cell aims to integrate additive manufacturing, automation, and post-processing into a single chain, designed to flexibly accommodate different components and production requirements while establishing conditions for repeatable and scalable processes.

Additionally, Siemens introduced an enhanced AM I Navigator tool to help companies identify and prioritize applications for Industrial AI in additive manufacturing. The company stated it will present the Agentic AI workflow for the humanoid robotic gripper at Formnext 2026 in Frankfurt.

출처:press.siemens.com

리서치

Report on German Youth Media Habits: Nearly Half Use AI for Search, One-Quarter Trust Answers Blindly

주제 · 德国青少年AI搜索信任중대
2026-11-16 20:08 GMT+8

According to the 'JIM-Studie 2026' published by the Southwest Media Education Research Association (mpfs), media habits among German teenagers aged 12 to 19 are shifting significantly, with artificial intelligence (AI) penetration standing out. The survey indicates that 63% of respondents use AI applications regularly, with information search becoming the most common purpose for the first time at 76%, up 6 percentage points from 2025. Overall, 47% of teens turn to chatbots when they need information, an increase of 8 percentage points year-over-year.

Critically, there is a notable variance in how young users evaluate AI-generated content. Data shows that 25% of respondents read only the AI answer and ignore other search results entirely, while another 19% click on at least one source cited within the AI response. This tendency to trust AI blindly is more pronounced among younger users: approximately one-third of those under 15 rely solely on AI answers, compared to 17% of 18- to 19-year-olds. This suggests that while older teens show greater awareness of the need to verify AI information, the overall risk remains significant.

Beyond search, AI is increasingly serving as a 'personal advisor' for some adolescents. About a quarter of AI users consult chatbots about health issues, and 15% discuss personal worries or psychological distress. In terms of role perception, 87% view AI as a search engine, but 46% see it as an advisor, 10% as a coach, and some even as a friend or therapist. This emotional projection has raised concerns among regulators about digital literacy, prompting calls to better guide teens in recognizing AI limitations and verifying information sources.

출처:heise online

리서치

OpenAI Donates £1M to The King’s Trust for UK Youth AI Skills Training

간단한 요점
2026-11-16 08:00 GMT+8

OpenAI has announced a partnership with The King’s Trust, a UK youth charity, including a £1 million donation aimed at helping up to 25,000 young people across the United Kingdom develop practical AI skills for the future workforce. The initiative prioritizes those who are not currently in education, employment, or training (NEET).

Spanning 18 months, the program forms part of a broader effort by The King’s Trust to launch nationally recognized AI qualifications. OpenAI stated that the collaboration addresses the rapid changes in the labor market driven by artificial intelligence. Tom Duff Gordon, VP & Head EMEA Policy at OpenAI, emphasized the goal of ensuring young people facing significant barriers have a real foothold in the AI economy.

The announcement coincides with The King’s Trust’s 50th anniversary, marking five decades of supporting more than 1.3 million young people. This partnership complements OpenAI’s existing “OpenAI Academy,” which provides free learning resources for students and job seekers.

출처:OpenAI

리서치

AI Geothermal Drilling Startup Bedrock Energy Raises $24M Series B to Target Data Center Cooling and Grid Capacity

중대
2026-11-16 21:00 GMT+8

Bedrock Energy, an AI-driven geothermal drilling startup, announced the completion of a $24 million Series B funding round led by Clean Energy Ventures and Clearvision Ventures. The company aims to significantly reduce heating and cooling costs for commercial buildings and data centers through its AI-guided drilling rigs.

Bedrock’s core technology combines directional drilling techniques from the oil and gas industry with proprietary subsurface models. The company claims this AI-directed approach can slash drilling times by up to 80%, thereby reducing costs. Traditional geothermal heat pumps have seen low adoption due to the high expense of drilling, a barrier Bedrock seeks to lower through improved efficiency and precision.

Beyond providing energy-efficient solutions for office buildings, Bedrock is actively engaging with data center developers. Some clients are interested in the technology’s “flat load profile”: deploying geothermal systems on neighboring properties can free up grid capacity, allowing data centers to operate earlier or increase available power. Joselyn Lai, co-founder and CEO of Bedrock, noted that a year of revenue for a data center is very meaningful.

Currently, Bedrock operates in Utah, Colorado, New York, Alberta, and Texas. The new funds will be used to add more rigs to its fleet and expand services into New England, New Jersey, the upper Midwest, and additional states in the West. Previously, the company raised a $12 million Series A in 2025 and an $8.5 million seed round in 2023.

출처:TechCrunch

리서치

Amazon-owned Zoox Launches Robotaxis on Uber App in Las Vegas, Marking First Third-Party Partnership

주제 · Zoox与Uber合作중대
2026-11-16 20:00 GMT+8

Amazon-owned autonomous vehicle company Zoox officially announced on Monday that its robotaxis are now available through the Uber app in Las Vegas. This marks Zoox's first partnership with a third-party ride-hailing platform, representing a significant milestone in its commercialization efforts.

For Uber, this integration is part of a broader strategy to become the primary provider of autonomous vehicle technology in North America, Europe, and the Middle East. Although Uber has partnered with more than 30 companies developing AV technology, not all collaborations have resulted in commercial deployments. This move allows Uber to offer another autonomous mobility option to its existing user base.

For Zoox, the partnership provides access to a larger pool of potential customers who might not otherwise try its service. Currently trailing market leaders Tesla and Waymo, this channel expansion offers Zoox a necessary boost. Initial limitations exist: customers cannot yet directly select a Zoox vehicle. Instead, users requesting specific ride types like UberX or Uber Comfort may be matched with a Zoox robotaxi. Riders can increase their chances by opting for autonomous rides in the Uber app settings.

Additionally, Zoox, Uber, and MGM Resorts International announced a multi-year partnership to improve pickup and drop-off locations and signage at MGM properties in Las Vegas. Earlier this year, the Nevada Transportation Authority approved a permit allowing Uber to deploy up to 1,000 robotaxis in Clark County through partnerships with Hyundai subsidiary Motional and Zoox.

The collaboration is expected to extend beyond Las Vegas. Zoox plans to begin initial operations in Los Angeles this year, with commercial service starting in 2027, also accessible via the Uber app. Uber CEO Dara Khosrowshahi stated that as availability and reliability improve, Uber plans to introduce a dedicated autonomous option in its app.

출처:TechCrunch

리서치

Alibaba's Qianwen Office Integrates Qwen4-Max and Third-Party Models, Opens User Selection

주제 · 千问办公模型选型중대
2026-11-16 19:44 GMT+8

Zhidx reported on November 16 that Alibaba's AI office product, "Qianwen Office," underwent a major upgrade by launching 10 large language models at once. The update includes not only Alibaba's newly released flagship model, Qwen4-Max, but also several third-party models such as Kimi-K3.1-Pro, GLM-5.3, DeepSeek-V4-Pro, MiniMax-M3.1-Flash, and Step-5-Preview. This signifies that Qianwen Office has officially opened the "model selection" feature to its users.

Previously, Qianwen Office primarily offered three fixed tiers: Standard, Advanced, and Qwen3.8-Max. With this upgrade, users can flexibly choose different models for tasks based on complexity, response speed requirements, and token cost budgets. For instance, Qwen4-Max is recommended for complex reasoning tasks; GLM-5.3 or DeepSeek-V4-Pro can be used for rule sorting, plan checking, and other analytical tasks; while lower-cost models like Qwen3.8-Flash, GLM-5.3-Flash, or DeepSeek-V4.1-Flash are suitable for high-frequency office tasks such as copywriting, information extraction, and content organization.

Notably, the newly added Qwen4-Max is the first model in Alibaba's Qwen4 series, released today. It features a new architecture with 2.4 trillion total parameters, activating 71 billion parameters per token, native visual understanding capabilities, and support for a 1 million token context window. According to evaluation results published by Alibaba, Qwen4-Max shows significant improvements over the previous Qwen3.8-Max across multiple professional Agent benchmarks. Specifically, in the Work360 benchmark, the average cost per single task dropped from 26.58 yuan to 11.16 yuan, a decrease of approximately 58%.

In addition to individual user choices, Qianwen Office has opened model management permissions to enterprise users. Enterprise administrators can configure model usage rights based on departments, user groups, or specific members, assigning corresponding models for different roles and tasks while centrally managing internal model resources. This move reflects a pragmatic trend in the current AI office sector: given that different models have advantages in reasoning capability, speed, and cost, integrating the strengths of various providers is becoming a more effective solution than relying on a single model.

출처:zhidx.com

리서치
다음 →
로그인 / 회원가입
독서리서치레이더투자 프레임워크