Reading

The demand for smart glasses is steadily growing, establishing them as a key trend in next-generation consumer technology. However, concerns are also rising regarding their potential use for secret recording or stalking in public places. On October 27, the tech media outlet Engadget published a practical guide aimed at general consumers. The core message is that while one can use physical features, wearer behavior, and Bluetooth detection apps to assist in judgment, no single method can confirm that secret recording is taking place; therefore, the appropriate response should be caution rather than confrontation.

Appearance offers the most intuitive clue. Most branded smart glasses feature small LED indicator lights that illuminate during recording. Additionally, because they must house electronic components, the temples (especially near the temples of the head) are typically thicker than those of ordinary glasses, and visible camera lenses or microphone grilles may be present on the frames. However, Engadget specifically notes that cheap counterfeit smart glasses without recording indicators can be purchased on e-commerce platforms such as TikTok Shop and Temu. These products completely bypass the first line of defense for visual identification.

When indicator lights are unreliable, behavioral observation can serve as a supplement. Actions such as repeatedly pressing the frame, appearing to read text displayed on the glasses, or muttering to oneself may signal that the wearer is operating the device. However, Engadget explicitly warns that these behaviors alone cannot be taken as definitive proof that recording is occurring.

Bluetooth detection apps represent a third path. Applications such as Nearby Glasses and ZuckOff, which have emerged recently, scan for nearby Bluetooth devices and compare them against common device identifiers used by AI glasses to issue alerts. However, these apps suffer from two fundamental flaws: they may fail to detect wearers who have turned off Bluetooth, and they may misidentify non-smart-glass devices as targets. Engadget advises using these apps only as an auxiliary confirmation tool and not relying on their results for judgment.

Regarding response strategies, Engadget suggests a tiered approach: do not suspect or confront someone simply because they are wearing AI glasses; if suspicious behavior is observed, maintain distance first; if an actual threat is perceived (such as being stalked), contact the police. For meetings or events involving confidential information, organizers can proactively clarify policies prohibiting the wearing of smart glasses. Engadget concludes that even if one suspects someone is secretly recording audio or video, one should remain calm and avoid rashly confronting the individual.

U.S. energy company GE Vernova Inc. (NYSE: GEV) announced on October 28, 2026, that it has signed an agreement to acquire Trench Group, a global leader in high-voltage grid equipment, from investment firm Triton Partners. The purchase price is approximately $9.0 billion, which is currently expected to be funded equally with cash and debt. The transaction is anticipated to close in late 2027, subject to customary regulatory approvals and consultation with employee representatives where legally required.

Trench Group has over 130 years of engineering heritage and serves more than 800 global utilities and hyperscalers with high- and medium-voltage bushings, instrument transformers, and coils. With more than 3,000 employees, its manufacturing footprint spans 11 sites across Europe, North America, and Asia. GE Vernova stated that the acquisition will directly enhance its two largest Electrification businesses, Grid Systems Integration and Power Transmission, by vertically integrating an existing supplier and increasing capacity for key grid equipment.

According to forecasts prepared by GE Vernova based on data provided by Trench Group, the acquired business expects $1.9 billion in revenue at an approximately 36% adjusted EBITDA margin in 2028. The $9.0 billion price tag reflects 13 times the 2028 adjusted EBITDA before synergies. Reuters reported that GE Vernova expects $50 million to $100 million in annualized cost synergies by 2030. Trench Group also holds approximately $5 billion in backlog and framework agreements.

GE Vernova stated in its press release that global electricity grids face surging demand from aging infrastructure, data centers, renewable energy, and industrial electrification, with higher voltage transmission lines increasingly becoming the backbone of the system, carrying up to six times more power per corridor than conventional lines. GE Vernova CEO Scott Strazik said the acquisition would help the company capitalize on a global high-voltage market expected to reach $130 billion by the end of the decade.

Woowa Brothers, the operator of South Korean food delivery platform Baemin, disclosed the operational results of its AI-based rider supply automation platform 'KAIROS' at the Woowa Tech Conference 2026 held in Seoul on October 28. According to data scientist Shin Sung-chan and data analyst Ju Ga-ram, KAIROS automates the entire process from rider supply prediction and target selection to operational response.

Traditional scheduling systems typically compare the number of riders needed for expected orders with the number of riders predicted to actually work. However, because each rider's operating time varies, it is difficult to accurately assess actual delivery processing capacity based solely on headcount. To address this, KAIROS changed its prediction basis from 'number of riders' to 'supply time'. The system utilizes approximately 20 machine learning models to analyze order demand, required operating time, and each rider's expected operating duration, considering the probability of riders responding to requests and their expected duration to select personnel.

This shift resulted in significant improvements in operational efficiency. According to Woowa Brothers, KAIROS reduced dispatch time by 25% and decreased excessively delayed deliveries by 50%. Additionally, to cope with sudden weather changes and order surges, the system shortened its prediction cycle from once daily to every 10 minutes, forecasting supply and demand conditions for the next 30 minutes to one hour. This led to a 20% reduction in regions deviating from quality standards, a 74% decrease in the duration of quality deterioration, and a 90% reduction in the operational team's response time.

At the DevDay Exchange event in Berlin on October 28, 2026, Romain Huet, OpenAI's head of developer experience, stated that the company's AI assistant Dots is expected to become more widely available in Europe "in a few weeks" following talks with regulators. Currently, personal Pro accounts in the European Economic Area, Switzerland, and the UK remain locked out of the service, though certain business accounts have gained access. This timeline marks the latest progress in OpenAI's push to deploy its core agent products in regulated markets.

Huet also disclosed regional adoption data: according to OpenAI, weekly active users of Codex, the company's coding tool, have grown twentyfold in Germany since January 2026. Although the company did not provide absolute numbers, it noted that nearly one in five desktop users taps skills for documents, presentations, or spreadsheets, indicating that the coding tool is being used for general office work. In June, OpenAI had reported a sevenfold increase based on its own figures.

Beyond product updates, the event highlighted OpenAI's intent to lock developers and businesses more tightly into its infrastructure. With the Agents API, OpenAI runs the agent infrastructure itself, using plugins and web apps to offer developers distribution channels. The company emphasized its 1.2 billion weekly ChatGPT users, aiming to replicate an app store model where developers build apps and OpenAI supplies the infrastructure and traffic. While the underlying runtime, Codex Harness, remains open source, developers relying on managed services and platform-specific integrations may find it difficult to switch ecosystems later.

Regarding the lack of a model supporting video inputs and the shutdown of the Sora video generator, OpenAI management attributed these decisions to high resource costs and weak customer demand. The company stated that customers primarily want coding models and agents, so video features were not prioritized, but did not rule out bringing them back later.

As power density in AI data centers increases, industry teams are exploring and validating 800 volt direct current (800 VDC) as a new power distribution method. DatacenterDynamics (DCD) published a podcast episode titled "Scaling Data Center Power with 800 VDC" on October 28, 2026, featuring contributors from Schneider Electric and NVIDIA who discuss how to design, protect, and test 800 VDC architectures as data center power scales.

The episode highlights several key learning points, including three 800 VDC architectural blueprints (A, B, and C) suited to different use cases, and a system-level approach to DC power distribution. Notably, it addresses personnel safety and arc-flash mitigation in 800 VDC systems, along with proof-of-concept testing on real equipment. According to the publisher's description, lab results showed that traditional circuit breakers cleared faults faster than initially estimated. This specific technical finding offers practical reference value for addressing safety constraints in high-voltage DC environments.

The discussion also covers fault detection, protection coordination, the challenge of defining system boundaries and interfaces, and how collaboration with the Open Compute Project (OCP) and IEEE supports alignment on objectives and standards. It is important to note that Schneider Electric sponsored the episode and contributed insights, so readers should consider potential vendor perspectives when evaluating the architectural recommendations.

Visa, the world's largest international card association, released its latest earnings on October 27, 2026, reporting that full-year revenue for fiscal 2026 (ending September 30) reached $46 billion, setting a new historical record. This performance was driven by sustained growth in consumer spending on travel, shopping, and dining, reflecting the resilience of global consumption momentum. According to Reuters, Visa's fourth-quarter revenue rose 13% year-over-year to $12.1 billion; adjusted earnings per share came in at $3.48, surpassing the market consensus of $3.44.

Transaction data showed that Visa's total volume increased 9% year-over-year in the fourth quarter, with transaction counts up 10%. Notably, cross-border transaction volume (excluding intra-Europe transactions) grew significantly by 14%, boosting interchange fee income. Visa Chief Financial Officer Chris Suh noted that spending growth remained robust across all consumer segments, with high-spending groups growing fastest and no obvious signs of weakness among lower-income consumers. Looking ahead to fiscal 2027, Visa projects full-year revenue growth of 10% to 12%, with adjusted EPS expected to grow 10% to 13%.

Despite both revenue and profit exceeding expectations, investor concerns over rising costs led to stock volatility. The report revealed that Visa's 'Client Incentives'—payments made to partner clients such as issuing banks—increased 21% year-over-year to $5.1 billion in the fourth quarter, higher than market forecasts. The rapid escalation of this key cost item raised fears that future profitability could be compressed. Consequently, Visa's shares fell approximately 1.3% in after-hours trading on October 27.

U.S. tech giant Meta has indicated that it may indefinitely delay the implementation of its controversial new business terms, originally scheduled to take effect on November 20th. This development comes amidst intense backlash from publishers concerned about the company's rights to scrape their content for AI training.

In a post on X, Meta communications head Andy Stone stated that the company does “not plan to make policy changes without a full consultation with industry groups and leaders.” Stone added, “We’ve heard publishers’ feedback on our proposed ads and business terms update... We’re committed to getting this right, and as we expect this consultation could extend beyond November 20, in which case we’d delay implementation of the updated terms. Until then our existing terms remain in place.”

The new business terms were initially set to go into effect on October 30th but appeared to grant Meta permission to freely scrape news sites using Facebook and Instagram for AI model training. Although Meta revised the language to remove the specific word “train,” legal experts told Nieman Lab that the update was insufficient to rule out such usage. Meta had previously delayed the effective date to November 20th to address these concerns.

Beyond opposition from the News/Media Alliance, lawmakers from both U.S. political parties have spoken out. Sen. Marsha Blackburn (R-TN) described the terms as an attempt to “exploit American creators and businesses,” while Sen. Adam Schiff (D-CA) urged Meta to reverse the changes. Additionally, digital advertisers and publishers in the European Union and the UK have called on local regulators to intervene and halt the implementation of the new terms.

Microsoft-owned code hosting platform GitHub announced at its GitHub Universe conference on October 28 that it is now processing 7.4 billion commits per month. This figure represents a significant jump from the 2.9 billion reported in August and far exceeds the 65 million monthly average seen in 2024. GitHub attributes this unprecedented growth primarily to "agentic demand," referring to code submissions generated automatically by AI programming agents.

To support this load, GitHub CTO Vlad Fedorov stated that the company has added 4 million CPU cores, mostly on Azure, along with 160 petabytes of high-speed storage and increased compute power for GitHub Actions. Network capacity was also expanded by 50 Tb/s. These hardware investments aim to alleviate recent service outages caused by traffic surges.

Beyond vertical scaling, GitHub reiterated its commitment to rearchitecting its platform to separate compute and storage. In an early October interview, GitHub COO Kyle Daigle revealed that the migration to Azure is roughly 60% complete. Internal benchmarks for the new architecture indicate up to 35x higher write throughput. This shift underscores that traditional monolithic code hosting architectures are hitting physical limits as AI agents become major participants in software development, necessitating reliance on hyperscale cloud infrastructure capabilities to maintain stability.

Taiwan-based memory manufacturer Macronix Electronics reported its third-quarter 2026 financial results on October 27, 2026. The company posted consolidated revenue of NT$24.237 billion for the quarter, representing a 27% increase from the previous quarter and a substantial 195% rise year-over-year. This performance marks the highest single-quarter revenue in the company's history.

Profitability saw a dramatic improvement, with operating gross margin surging to 70.5%, up from 64.4% in the previous quarter and 13.5% in the same period last year. Approximately NT$750 million of this margin boost was attributed to inventory write-back gains. Operating net margin reached 53.4%, while net income after tax hit NT$10.951 billion, a 1,370% year-over-year increase, resulting in basic earnings per share (EPS) of NT$5.52.

The primary drivers were strong demand for high-density NOR Flash in AI data centers and severe supply shortages in the SLC NAND and eMMC markets. In Q3, NAND Flash (including eMMC) accounted for 47% of total revenue, growing 768% year-over-year. NOR Flash contributed 46% of revenue, up 158% year-over-year, with computer and data center applications making up 47% of NOR Flash sales.

Macronix Chairman Wu Min-chiu stated that limited semiconductor capacity and the long lead time for new equipment investments mean short-term growth relies primarily on price increases and product mix optimization. He highlighted that beyond AI data centers, the rise of satellite and space communications is rapidly boosting demand for eMMC and Flash, positioning the company favorably in this emerging segment.

At the ElevenLabs Summit Bengaluru, Rishabhdhwaj Singh, SVP of Engineering at Indian home services platform Urban Company, played real production call recordings featuring AI relationship managers and AI trainers built on top of ElevenLabs voice models. These demonstrations addressed core pain points in managing large-scale service professional networks.

Urban Company connects thousands of service professionals with millions of customers across India, the UAE, and Singapore. Unable to assign a dedicated human relationship manager to every professional, the team developed AI solutions. The demo included guiding a first-time professional to a customer's home, rescheduling bookings mid-call via backend tool integration, and collecting referrals immediately after reversing a cancellation.

Beyond external relationship management, Urban Company deployed AI trainers to guide technicians through diagnosing faulty water purifier pre-filters and conducting best-practice training sessions, with some calls lasting up to 30 minutes. According to Singh, the platform now handles close to 5 million minutes of such AI calls per month.

Data indicates that the AI relationship manager independently resolves 80% of queries, routing the remaining 20% to humans, and collects most referrals in cities where it is live. Singh also explained why the trainer and the relationship manager use different voices and personas from the ElevenLabs voice library to suit their respective roles and user expectations.

According to Nikkei Asia, major personal computer manufacturers HP, Asus, and Acer are moving some of their manufacturing capacity back to China as skyrocketing memory and storage chip prices, driven by artificial intelligence demand, squeeze their margins. This shift aims to leverage China's mature supply chain ecosystem and relatively lower production costs to offset rising bill-of-materials (BOM) expenses.

Notably, these brands are not returning to traditional contract manufacturers like Foxconn, which have pivoted their capacity toward AI data center servers. Instead, they are signing contracts with smaller Chinese contract manufacturers such as Luxshare, Huaqin Technology, and Longcheer Technology. These emerging players are seeking to establish themselves in PC manufacturing and claim better access to domestic memory chips from suppliers like CXMT.

Previously, many PC brands gradually moved production out of China in 2024 and 2025 to adopt a 'China+1' diversification strategy amid U.S. sanctions and tariffs. However, with AI-induced memory prices climbing 500% in 12 months and costs for other components like PCBs and fiberglass cloth also rising, the economic case for maintaining dispersed production has weakened significantly.

This reversal is impacting newly established supply chains in Southeast Asia. A manager at a PC component supplier revealed that two new production lines built in Thailand for PC clients are now idle as substantial orders move back to China, with no clear timeline for recovery. This suggests that despite ongoing geopolitical risks, immediate economic realities are reshaping the geographic landscape of hardware manufacturing.

Former Yahoo CEO and early Google engineer Marissa Mayer recently found herself in a public relations storm due to an interview. While introducing her newly developed AI assistant, Dazzle, she showcased a feature she considered 'impressive': the AI, which learns from user photo albums, generated three descriptions for her ten-year-old daughter’s school birthday presentation—'empathetic, creative, and loves animals.'

This description resembles a generic horoscope, likely applicable to any ten-year-old child. Progressive activist Leah Greenberg commented that Mayer voluntarily showing this to reporters 'perfectly illustrates how deeply the culture of Silicon Valley has diverged from the rest of humanity.'

However, Silberling’s argument goes beyond mocking a CEO’s PR misstep. She points out that Mayer triggered such significant backlash because the public has been repeatedly conditioned to frustration—from OpenAI CEO Sam Altman stating last year on the Jimmy Fallon show that he 'cannot imagine raising a newborn without ChatGPT,' to his claim months ago on X that a 'cool use case' for ChatGPT Work was generating a podcast to 'explain your child’s interests' (which drew numerous replies asking, 'Why don’t you just talk to your kids?'), to Meta CEO Mark Zuckerberg predicting that outsourcing emotional needs to AI therapists and AI romantic partners will become more socially acceptable.

Silberling suggests that part of the reason behind these executives’ remarks may lie in the tech industry’s efficiency-centric culture—where people optimize sleep with health trackers and inject unregulated 'peptide stacks' to maximize productivity. If you are building electric vehicles or chips, efficiency is indeed a clear direction. But human relationships are not a tech product required to be delivered with minimal capital as quickly as possible.

She cites data from the Pew Research Center as an anchor: 50% of U.S. adults believe AI will weaken people’s ability to form meaningful relationships, while only 5% think AI will make these connections easier. This indicates a huge gap between the 'frictionless future' envisioned by executives and the intuitive judgment of the majority of the public—yet executives tend to view opponents as a small, angry fringe group.

Silberling concludes that human relationships are inherently messy, but it is precisely the investment within this chaos that gives life meaning. When we delegate these interactions to computers, we surrender not only efficiency but also the difficult yet fulfilling experience of being human.

Mayer later defended herself on X, claiming she was merely testing prototype capabilities rather than outsourcing parenting—Dazzle did retrieve photos of 'forgotten kindergarten projects' from the album, such as a 'snake devenomer' and a 'sandcat playground,' showcasing her daughter’s creativity. However, in the same interview, she did propose using AI to plan children’s birthday party themes, arrange playdates, decide on Halloween costumes, and even have her AI assistant coordinate plans with her husband’s AI assistant. When asked if imagining this 'frictionless future planned by AI' was depressing, she replied, 'I haven’t really thought about that.'