Techin' Geek

A reported Bloomberg article referencing Netflix data indicates a growing trend of viewers walking away from beloved shows before they reach their sophomore season. The probable causes are apparent: Netflix often discontinues series, there are prolonged gaps between seasons, and a significant portion of Netflix’s offerings is tailored to algorithms rather than artistic expression.

Moreover, the data highlights a transformation in how audiences engage with entertainment. Netflix’s hallmark innovation – binge-watching – was conceived during a time when streaming was battling against conventional television. Currently, however, Netflix is in competition with TikTok, YouTube, Reels, and various microdrama applications. This evolution renders Netflix’s binge strategy feel like an antiquated concept from a bygone era.

Binge-watching enabled Netflix to surpass traditional TV

The release of an entire season of “House of Cards” by Netflix in February 2013 was revolutionary.
With ad-free, internet-based programming, we were liberated from the weekly schedule of episodes interrupted by advertisements. Binge-worthy shows allowed audiences to enjoy hours of entertainment, quickly forming connections with the content and characters that would have traditionally taken years to foster. Additionally, viewers could access these shows at any time, rather than just when networks decided to air them, as was the case with traditional television.

This consumption pattern made sense when Netflix was primarily competing against conventional television like broadcast, cable, and satellite. But Netflix triumphed in that contest. In June 2025, Nielsen revealed that the streaming format akin to Netflix had, for the first time, outperformed broadcast and cable viewing, marking a clear indication that Netflix’s original competition was no longer a concern.

Now, Netflix’s rivals are not the outdated TV models but the modern iterations: video applications.

TikTok and YouTube represent today’s challenges

With the ascent of TikTok, Reels, and other short-form video platforms, there’s little incentive to use Netflix when you have spare hours to fill with effortless entertainment. There’s an infinite, complimentary supply of videos available as an alternative.

eMarketer analysts noted that TikTok was already close to rivaling Netflix concerning time spent in 2024, with U.S. adults averaging 62.1 minutes per day streaming on Netflix and 58.4 minutes on TikTok. Furthermore, in 2024, the Financial Times reported that TikTok users globally averaged 95 minutes per day on the app, the highest engagement rate among major social networks.

Additionally, YouTube combines both short and longer forms of content. A report from Digital i in the current year indicated that YouTube exceeded Netflix in average daily viewing for the first time, with 99.1 minutes daily in 2025 compared to Netflix’s 93.4 minutes.

These market reports utilize varied methodologies and demographics, thus they should be viewed with caution — however, they trend in the same direction. YouTube and platforms like TikTok are indeed Netflix’s primary competitors, not traditional television.

Netflix has even recognized this existential challenge via a product redesign in April that introduced a TikTok-esque feed based on Netflix offerings.

Where Netflix misjudges the feed is in its presentation as a means to assist you in discovering what to watch, rather than being content worth viewing. It’s understandable why Netflix chose this path, considering its content library, but it may not align with the preferences of users. Nowadays, many individuals with fleeting attention spans are increasingly gravitating towards microdrama applications when they seek serialized narratives that can be consumed within minutes.

Data from the app intelligence company Appfigures revealed that one top microdrama application, ReelShort, achieved approximately $1.2 billion in gross consumer spending in 2025, up 119% from 2024, as reported by TechCrunch’s Amanda Silberling. In contrast, another prominent app, DramaBox, generated $276 million in gross consumer spending last year, more than doubling its figures from 2024. Even TikTok has acknowledged the rivalry by introducing its own microdrama app to assess market interest in such content.

What’s next for Netflix?

What position does this leave Netflix in, whose hallmark has been the release of full seasons for swift viewing?

It will likely need to reconsider how it greenlights, develops, and launches what it terms a “TV show.”

This doesn’t imply that Netflix’s model must completely shift to short-form to stay competitive, but it may require reevaluating how consumers prefer to stream. Viewers might no longer wish to dedicate the hours and weeks required to complete a show and its subsequent seasons. They are looking for content that feels more “completable,” similar to how one can finish a YouTube video or a TikTok series from a creator.

An easy adjustment might have Netflix focusing on single-season shows, typically referred to as miniseries or limited series, which would enable audiences to engage with a finished product without the anxiety of unfinished cliffhangers and cancellation uncertainty.

Moreover, Netflix could look into splitting shows into smaller segments, akin to the ahead-of-its-time Quibi model.

The Jeffrey Katzenberg-backed startup, Quibi, had anticipated that consumers would eventually lean towards TV content designed for brief viewing sessions. Unfortunately for Quibi, the pandemic disrupted this notion, as audiences suddenly had abundant time to watch TV, leading to its downfall.

Numerous Netflix programs could be easily adapted for shorter viewing experiences, especially lighter competition-centric shows like “Nailed It,” “Is It Cake?,” or “Squid Game: The Challenge.” Concurrently, Netflix could undoubtedly craft superior microdramas compared to the current offerings, which often suffer from poor acting and absurd narratives.

To drum up interest in its higher-caliber content, certain Netflix shows could shift to a weekly release approach. This has already proven effective in some cases, such as the planned weekly episode drops of its reality series “Love Is Blind,” creating buzz as viewers engage with the episodes simultaneously. (Quicker consumption models could also be successful, like Peacock’s “Love Island USA,” a reality sensation of the summer, with nearly daily new episodes).

However, rather than exploring various forms of short content for quick entertainment paired with slower season releases, or concentrating more on watchable miniseries, Netflix has been experimenting in different directions.

Recently, it has broadened its offerings with podcasts, which reportedly have low viewership, and live content, which can be unpredictable. Regarding the latter, Netflix’s investments in live sports have generally performed well, yet its recent foray into live reality competition shows, “Star Search,” has already been canceled despite a clever real-time voting feature. This area still requires improvement.

Bloomberg’s report characterized Netflix’s challenge as a shortcoming in cultivating dedicated TV viewers who would return for a Season 2, but the fundamental issue confronting the streaming service is considerably more extensive. Netflix may need to reconsider whether it should continue focusing on competing with traditional television and its established series or pivot towards entertainment projects characterized by tighter storytelling arcs that conclude more swiftly.

To strike the right balance between audiences abandoning cable and those still seeking a superior alternative to TikTok, Netflix finds itself in need of a reinvention of television once more.

In the previous week, experts from the cloud security company Sysdig reported they had identified the initial documented instance of “agentic ransomware.” This extortion scheme, named JadePuffer, involved an AI agent — as opposed to a human — managing the entire technical execution of a cyberattack in the real world. The agent infiltrated a vulnerable server, acquired credentials, navigated through the target’s network, encrypted files, and even drafted its own ransom note, adjusting to hurdles much like a human hacker would. Coverage of the funding mentioned that it was conducted “without any human oversight,” asserting “no human at the keyboard.”

However, that’s not entirely the full story. During an interview on Monday with CyberScoop, Sysdig’s Michael Clark, the firm’s senior director of threat research, emphasized that a human was indeed involved — just not in the technical execution aspects. “A human still set up and directed the operation and provided the necessary infrastructure behind it, including the command-and-control server, the staging server for the stolen data, and selecting a victim,” Clark explained. The credentials utilized to access the victim’s database were not obtained by the AI agent directly; they were acquired separately by someone through a previous breach and provided to the operation.

None of this contradicts Sysdig’s initial assertion, with the technical specifics of the attack being remarkable in themselves — even astonishing. The agent gained access via a known vulnerability in Langflow, a widely used open-source tool for creating LLM applications, and subsequently targeted a production MySQL server, exploiting another recognized flaw to obtain admin privileges. It encrypted more than 1,300 configuration records and not only composed a ransom note itself but also included a Bitcoin address for payment. Sysdig has not disclosed the identity of the targeted entity.

The methods utilized seem rather typical, yet the rapidity and clarity exhibited were noteworthy. The agent resolved a failed login in just 31 seconds, detailing its reasoning through natural-language code comments throughout the process.

A detail that initially appeared to obfuscate the narrative has been clarified. Clark had informed CyberScoop that Sysdig found “multiple models were used in the attack,” referencing harvested keys from OpenAI, Anthropic, DeepSeek, and Gemini — phrasing that left the door open regarding whether various models were simultaneously involved at different stages of the breach. When prompted for clarification, Clark told TechCrunch that those keys were merely part of what the agent pilfered, not indicators of what was operating it.

“The agent scoured the Langflow host for anything of value — provider API keys, cloud credentials, cryptocurrency wallets, and database configurations — and those provider keys were part of the bounty,” he stated via email. “They reflect what the attacker deemed worthwhile to take, yet they do not convey which model was making the decisions.”

Regarding the model specifically powering JadePuffer, Clark noted Sysdig “was unable to determine the specific model operating the agent” and lacks insight into its system prompt or setup.

The theory from Microsoft researcher Geoff McDonald, shared on LinkedIn a few days prior, is worth reconsidering in this context. McDonald speculated that an open-weight model with safety training removed, rather than a cutting-edge model, was behind the attack, based on his own red-teaming experiences indicating that safety layers in frontier labs perform effectively. Sysdig’s account does not confirm nor deny this possibility.

McDonald’s post also cautioned that ransomware operations are increasingly limited by the attacker’s budget rather than human labor, suggesting the potential for “thousands or tens of thousands of simultaneous campaigns.” This concern is somewhat challenging to reconcile with what Clark outlined on Monday. (If a human must still select each victim, arrange infrastructure, and secure database credentials for every operation, that poses somewhat of a bottleneck.)

Regardless, Clark informed CyberScoop that while Sysdig has not observed the same operation target other victims so far, he anticipates changes soon due to the low cost of operating an agent.

SK Hynix, a memory chip producer from South Korea and a competitor to Samsung and the U.S.-based Micron, is set to offer nearly 17.8 million shares in an IPO in the U.S., according to a statement made by the company on Monday. If the shares perform well (which appears likely), the company could potentially secure around $28 billion, based on the share price of SK Hynix at the close last Friday in Seoul, as reported by Bloomberg.

The company will issue American depositary receipts (ADRs), which are certificates allowing U.S. investors to acquire foreign stocks without engaging in trading on international exchanges directly. Each ADR will correspond to one-tenth of a common share. Pricing for these securities is projected for Thursday, with trading anticipated to begin on Friday.

Similar to Micron, SK Hynix is benefiting from a surge driven by AI, reflected in both sales figures and stock valuation. They reported almost a 200% increase in revenues for the first quarter compared to the same quarter from the previous year, and their stock has risen about 260% thus far in 2023. This is due to the high memory demands of AI operational systems. As major companies like Amazon, Microsoft, Google, and Oracle strive to develop AI facilities and as new AI data centers proliferate across the country, demand is surpassing supply, leading to a shortage of memory chips — including high-bandwidth memory (HBM), DRAM, and NAND (which are the various types of chips that are utilized for data storage and transfer within AI systems). This scenario has been labeled as “RAMageddon.” Executives at Apple have indicated that the shortage is prompting them to increase prices on Mac computers and iPads.

South Korean tech firms, spearheaded by SK Hynix and Samsung, have committed to invest over $550 billion in developing new manufacturing capabilities to meet the rising demand. However, this investment is considered risky. By the time these facilities are operational, the memory requirements for AI may evolve, potentially resulting in excess supply and a subsequent drop in prices. Nonetheless, currently, Wall Street is on the lookout for another Nvidia, and memory chip manufacturers are among the most viable candidates available.

Micron, the most comparable U.S. entity, has soared nearly 700% within the past year, attaining a valuation exceeding $1 trillion, driven by unprecedented memory demand and revenue propelled by AI.

Vercel, recognized for its cloud infrastructure enabling developers to deploy agents without server management, has discreetly established itself as a pivotal player in AI software. The company observes 6 million deployments daily, with half initiated by coding agents, and over 1 trillion tokens processed through its AI gateway each day.

Following the company’s ShipNYC conference last week, we engaged in a discussion with Vercel CEO Guillermo Rauch about the current AI landscape and how platform companies like Vercel find themselves in competition with major research laboratories. Below is a lightly edited transcript.

This year seems to bring a fresh energy in the community, with fewer pilot projects and a greater emphasis on practical implementation. I suspect you’ve encountered that with clients, but I’m interested in how that journey has unfolded internally at Vercel.

Last year focused on prototyping. The potential was limitless, release the agents, everyone can create, and so forth. We accomplished that, gaining valuable insights from having numerous agents organically created and deployed within the organization, only to later confront the actualities of agents in production and their associated challenges.

The most significant takeaway for me was identifying the major use cases, the two standout applications of agents. The first is, of course, the coding agent. This is a key driver of token usage globally, but with such a high volume of software production, there must be a repository. The second major application of agents is the internal agent that facilitates company operations. The main issue there is securing data access, auditing agent activities, and establishing a history of all tool interactions and access permissions that the agent had to navigate for task completion.

To address this, we developed a framework named Eve, which allows you to outline an agent’s directives and abilities in natural language. Another tool is Vercel Sandbox, which confines the agent in a controlled environment. While it can still express its intelligence, we can enforce policies regarding what data it can access and what data can exit the sandbox.

What kinds of issues does this help you avoid?

The biggest benefit of the sandbox is data management. A significant risk of AI that I consistently consider is when you utilize a coding IDE like Devin or Cursor; if you’re in an inappropriate environment, they might train on your entire codebase. I remember discussing this with the president of Airbus. They possess decades of highly specialized C++ code for aerospace engineering. If someone mistakenly installs the wrong developer tool, all the code can be sent to the cloud for training.

I’m interested in learning more about that second primary use case. We are familiar with coding agents, but what does an internal corporate agent look like in reality?

For example, there’s a sales representative at Vercel. Her role is to expand existing accounts. The bottleneck for individuals like her hasn’t been a lack of creativity, intelligence, or relationship-building skills; it’s been data. “I need to know which accounts are growing rapidly. Provide me with the five accounts that have increased the most seats in the past fortnight so I can prioritize.” In the past, she couldn’t ask that question and had to wait for a Q1 project for a new sales dashboard to conclude.

We faced that bottleneck for years at Vercel, which was quite frustrating because, on the R&D side, we are the most agile company globally. However, on the sales front, the Salesforce engineering aspect was something I was utterly unprepared for. I had never used Salesforce before I started.

Now, I believe I can truly impact the entire organization because Eve can be utilized for our customer-serving agents and enhance productivity. The same technology simply utilizes APIs. Agents are compelling businesses to become more open, leading to significant long-term consequences. Many of these SaaS giants have constructed their empires by entraping your data, which is incompatible with agents.

How do you perceive client relationships with the large AI labs evolving?

Last year, many were selecting a singular lab partner, committing to building everything on OpenAI or Anthropic. Now, they’re realizing how it all fits together — model, harness, data platform, sandbox, gateway — everything is modular. You can employ OpenAI, Anthropic, or Gemini. We are witnessing significant growth in Gemini, even if it isn’t prominently featured in the news, because companies are now prioritizing production. The truth is, when optimizing for production, you start considering price/performance, and Gemini models exhibit excellent price/performance metrics. Open models are also gaining traction, with DeepSeek and GLM-5.2 becoming increasingly popular. The data speaks for itself.

There are areas where you are in direct competition with the labs as well, correct? Just recently, OpenAI unveiled a new suite of tools that enables direct web publishing without leaving the OpenAI ecosystem.

It’s a logical progression for them to host small websites. It creates a fantastic opportunity for us because individuals will begin to view ChatGPT as a platform for website creation. If they persist in querying the model about web hosting, it may recommend our services. However, you are correct that as models and platforms enhance their capabilities, they directly compete with the existing infrastructure platforms.

I genuinely believe we are at a crossroads regarding whether the model and the agent will be interconnected.

Will you derive all your intelligence from one source? Or will you receive a module, library, or building block from one provider and then build upon it? This aligns more with traditional software engineering, which is precisely what we’re introducing to the market. We aspire to be the AWS of this era, thus we are striving for a landscape of open protocols.

On Monday, Microsoft announced it has cut approximately 4,800 positions, representing 2.1% of its global staff, contributing to the wave of AI-driven layoffs impacting the tech sector. The company stated that the positions being eliminated are “not being replaced by AI,” while also admitting that “AI is altering work processes” and automating numerous daily tasks.

These reductions are part of what many in the tech industry perceive as a crisis: firms reporting record earnings while simultaneously downsizing their workforces, indicating AI as both the catalyst for growth and the justification for these cuts. According to outplacement company Challenger, Gray & Christmas, tech layoffs reached their highest monthly count in years in May, with AI cited as the leading cause. In 2026, almost 120,000 tech jobs have been eliminated, based on data from Layoffs.fyi, a tracker monitoring layoffs in the industry since 2020.

We recently discussed why this reasoning could be something companies should reconsider, particularly since many of the teams they’re now reducing expanded significantly during the pandemic hiring boom, prompting inquiries about the current situation. Below is a chronological overview — in reverse order — of major tech firms that have reported significant layoffs this year attributing them to AI.

---

Oracle — June 22, 2026. Oracle revealed at the end of June that it had cut its workforce by 21,000 employees over the past year, a decrease of 13%, indicating greater reductions than previously acknowledged, including those related to AI. “The integration and utilization of AI technologies in our operations have led, and may continue to lead, to workforce reductions,” the firm stated in an annual regulatory financial report.

GitLab — June 3, 2026. GitLab laid off around 350 employees, approximately 14% of its personnel, to invest in AI infrastructure and manage increased traffic from AI workflows. CEO Bill Staples mentioned that agentic workloads are “driving competitors to their limits” and that the company has embarked on a “generational rebuild” of its core infrastructure to meet what he described as 100x growth demands. GitLab is withdrawing from 22 countries, streamlining management structures, and collaborating with an unnamed AI lab to enhance its platform for large-scale workloads. The company reported first-quarter earnings of $264 million, a 23% increase year-over-year, and anticipates restructuring costs between $30 and $35 million.

Google — ongoing through May. Alphabet’s Google has subtly reduced staff within its Cloud division, including members of its Threat Intelligence Group and Mandiant-related cybersecurity personnel, despite a 63% growth in Cloud revenue, surpassing $20 billion for the first time, and nearly doubling its backlog to over $460 billion. Over the past year, Google has cut more than a third of the managers supervising small teams — a 35% reduction in managers with fewer direct reports. Unlike the majority of firms on this list, Google has not disclosed a specific overall number of layoffs — cuts have been made through an ongoing performance review process, a voluntary buyout scheme, and organizational changes, with external estimates placing the 2026 total at between 1,500 and over 3,000 engineers.

Intuit — May 20, 2026. Intuit declared intentions to eliminate around 3,000 jobs — about 17% of its entire workforce — during a restructuring focused on reducing complexity and reallocating resources towards AI. CEO Sasan Goodarzi reportedly communicated to staff that the company aims to simplify its structure to enhance product delivery.

Meta — May 20-21, 2026. Meta terminated approximately 8,000 workers, about 10% of its total staff, while reallocating around 7,000 employees into new AI-centered roles (which they reportedly dislike). CEO Mark Zuckerberg informed teams that the layoffs were essential because “success isn’t guaranteed” in AI.

Cisco — May 14, 2026. Cisco announced plans to eliminate nearly 4,000 jobs, equivalent to about 5% of its workforce, even while posting better-than-anticipated profit and revenue figures. CFO Mark Patterson stated: “This wasn’t purely a savings-driven restructuring… it’s more [about] realigning … resources related to silicon, optics, security, and AI.”

Cloudflare — May 7-8, 2026. Cloudflare reduced about 20% of its workforce (1,100 individuals), reporting quarterly earnings of $639.8 million, a 34% increase year-over-year, marking the highest quarter in the company’s history. CEO Matthew Prince indicated that “the vast majority of those we laid off last week were measures” — middle management, finance, legal, internal auditing, and revenue recognition.

General Motors — May 12, 2026. GM cut between 500 and 600 jobs, primarily in IT positions located in Austin, Texas, and Warren, Michigan, citing a reevaluation of workforce needs amidst unpredictable market conditions. A source familiar with the layoffs informed CNBC that AI influenced the decision, though it wasn’t the sole factor. GM’s comment stated that it is “transforming its Information Technology structure to better prepare the company for the future.” Despite the reductions, the company still has roughly 80 IT roles open, including positions in AI, motorsports, and autonomous vehicle development.

Coinbase — May 5, 2026. The cryptocurrency platform announced it is laying off about 700 staff, or 14% of its workforce, as part of a restructuring focused on addressing market fluctuations and enhancing AI efficiency. The organization has streamlined its structure to five layers below the CEO and COO, and stated it would experiment with “one-person teams” uniting engineering, design, and product functions. CEO Brian Armstrong wrote that AI has significantly accelerated work — “engineers utilize AI to accomplish in days what formerly took a team weeks” — and emphasized the necessity to “integrate AI into every aspect of our roles.”

PayPal — May 5, 2026. PayPal revealed plans to cut approximately 20% of its workforce over the next two to three years — exceeding 4,500 jobs — as part of a turnaround strategy focused on the adoption of AI and simplifying the organization. CEO Enrique Lores told investors the company would “vigorously incorporate AI” in its development procedures and established a new “AI transformation and simplification” team directly reporting to him, charged with redesigning the company’s processes “function by function.” Lores characterized the layoffs as a reduction of organizational layers and noted that AI will extend beyond coding to customer service, support operations, and risk management.

Microsoft — April-May 2026. Microsoft provided voluntary buyouts without revealing how many employees would be affected. CFO Amy Hood mentioned that the overall headcount decreased year-over-year in fiscal Q3, and anticipates continuing to decline as the firm emphasizes “creating high-performing teams that work with speed and agility” amid increasing AI investments.

Snap — April 16, 2026. Snap eliminated around 16% of its global workforce — close to 1,000 full-time employees — and shut down over 300 open positions, with CEO Evan Spiegel attributing advancements in AI as a significant factor. “Accelerated progress in artificial intelligence enables our teams to diminish repetitive tasks, enhance speed, and better serve our community, partners, and advertisers,” Spiegel stated in a memo filed with the SEC. The company noted it has already observed small teams utilizing AI tools to foster progress across Snapchat+, advertisement platform performance, and operational efficiency.

IBM — ongoing through 2026. Cuts in Q4 2025 and April 2026 Red Hat engineering reductions yield estimates of 3,000 to 9,000 U.S. positions eliminated, raising IBM’s cumulative total since September 2024 above 15,000. Bloomberg reported IBM intends to triple its U.S. entry-level hiring for AI and hybrid-cloud positions, even as around 200 HR positions were supplanted by AI agents. An IBM representative described the Q4 2025 layoffs as a regular rebalancing impacting “a low single-digit percentage” of its global staff.

Atlassian — March 11, 2026. Atlassian reduced approximately 1,600 jobs (10% of its workforce) to “rebalance” towards AI and enterprise sales, even as shares rose nearly 2% following the announcement. CEO Mike Cannon-Brookes stated: “Our approach is not ‘AI replaces people.’ However, it would be misleading to pretend AI does not alter the skill mix we require or the number of roles necessary in specific fields. It does.”

Dell — January 30 (disclosed in March 2026). Dell’s total workforce decreased by about 10% in fiscal 2026 — around 11,000 jobs — down to approximately 97,000 employees from 108,000 a year prior, with $569 million allocated for severance. The reductions occurred as Dell predicted its AI-optimized server revenue could double in fiscal 2027.

Oracle — March 5-31, 2026. As mentioned before, Oracle started informing workers of impending job cuts through terminal emails. These cuts were announced despite Oracle reporting $3.7 billion in quarterly net income, a 27% year-over-year increase, with remaining performance obligations rising 325% to $553 billion — savings redirected towards AI data centers. The total cuts would later reach 21,000 over the course of 12 months, which Oracle disclosed in its June 22 annual report.

Block — February 26-27, 2026. Jack Dorsey’s Block eliminated 4,000 jobs — nearly half its workforce, reducing numbers to under 6,000 from over 10,000. Dorsey posted on X: “We are already witnessing that the intelligence tools we are developing and using, combined with smaller and flatter teams, are creating a new approach to work that fundamentally transforms what it means to build and operate a company.” He added: “I believe most companies are behind the curve. Within the next year, I predict that the majority will reach the same realization and implement similar organizational changes.”

Salesforce — February 10, 2026. Salesforce laid off fewer than 1,000 staff across marketing, product management, data analysis, and its Agentforce AI division. The firm informed Fortune, “Thanks to the advantages and efficiencies of Agentforce, we’ve observed a decline in the support cases we handle, and we no longer need to actively replace support engineer roles.” This was preceded by an earlier cut of about 4,000 customer-support positions, reducing the team from roughly 9,000 to 5,000, with CEO Marc Benioff stating that the firm required “fewer individuals” since AI agents manage the workload.

Amazon — January 28, 2026. Amazon eliminated 16,000 corporate roles, following 14,000 cuts in October 2025 — around 9% of its corporate workforce in just three months. The company indicated this was part of “strengthen[ing] our organization by minimizing layers, enhancing ownership, and removing bureaucracy.” CEO Andy Jassy remarked in June 2025 that, “As we introduce more generative AI and agents, it should alter the way we carry out our work. We will require fewer people to perform some of the tasks currently being done… over the next few years, we anticipate this will decrease our total corporate workforce as we achieve efficiency gains from thorough AI usage across the organization.”

For those seeking an e-reader capable of purchasing pricey bestsellers from their nearby independent bookstores, the search will inevitably lead to Rakuten’s Kobo eReader. However, this reputation is predominantly founded on obsolete methods for acquiring Kobo titles, which require creating a Kobo account via the websites of the bookstores.

Although a few independent bookstores across the country still maintain Kobo support through this online route, most of that support has diminished over the years. I have found it impossible to obtain e-books for my newly acquired Kobo from any of my cherished local bookstores.

I aim to support these bookstores and prefer not to possess a physical copy of every book I read. Additionally, I favor using an e-reader over an Android or iOS application on a phone or tablet, as e-readers provide extended battery life, digital ink, and anti-glare screens. These features allow me to read comfortably for long periods, even outside, much like a physical book.

One potential solution for Kobo users, which was initially expected in 2025, involved a collaboration with Bookshop.org, an alternative to Amazon that supports local bookstores with each transaction. Bookshop.org currently provides e-books through its mobile application for iOS and Android devices.

That partnership was initially scheduled for 2025, then postponed to 2026, and briefly appeared earlier this year as though it might be delayed indefinitely.

After Bookshop.org modified the language on its webpage regarding Kobo support, changing “2026” to “sometime in the future,” I reached out for a status update.

Bookshop.org’s founder and CEO, Andy Hunter, informed me in an email response that progress with Kobo has now been achieved. The webpage has been revised, indicating that support is anticipated to be launched “later this year.”

“The Kobo integration is something both Kobo and Bookshop.org want to see happen,” Hunter stated.

The delay has stemmed from both business and technical aspects to ensure it is “executed in a manner that respects publisher demands for digital rights management. It took us some time to finalize the business arrangements and allocate the necessary engineering resources,” he clarified.

Hunter, whose company also competes with Amazon by offering physical books, stated that his engineers have been focused on enhancing the mobile app, which was launched approximately 15 months ago.

Their focus is now shifting back to Kobo support, though the timing remains uncertain. “We have recently agreed on business terms with Kobo, and we are confident that the collaboration will occur, but we cannot guarantee a specific launch date until the engineering work is more advanced,” Hunter noted.

Clearly, regardless of whether Bookshop.org successfully establishes Kobo support, Kobo users are not required to acquire all their books from the Japanese-based e-commerce giant Rakuten. Kobo users can access a wide array of digital rights management (DRM)-free titles on their readers, along with a substantial selection (though not all) of library books provided through Overdrive. The independent e-book retailer eBooks.com similarly offers DRM-protected books in a format compatible with Kobo, as they claim.

Another alternative, if your objective is to assist local bookstores through e-book purchases, is to opt for a different e-reader. An Android reader like Boox or Meebook that supports the Google Play app store should be able to download the Bookshop.org app, says the bookseller.

Nevertheless, like many other Kobo owners, I am hopeful for the Bookshop.org integration to come to fruition. Supporting local independent bookstores was my primary motivation for choosing this specific e-reader, as I was misled by my online research (and the assured guidance of ChatGPT).

Now that I have a Kobo Libra Colour, I truly appreciate its reading display, quick response time, and prolonged battery life. I also continue to keep my six-year-old Kindle for the same reasons.

Yet, I have a deep affection for local small-business bookstores, with their personalized suggestions, promotion of local authors, and genuine passion for literature. Here’s hoping that the leading e-commerce platform that champions local shops, Bookshop.org, will soon provide support for the widely used Kobo device, which boasts 12 million users across 190 countries.

Take this as a late notice: A new alteration to Google’s privacy controls permits the company to retain more of your information, encompassing media like “images, files, and audio and video recordings,” to enhance its AI models. Essentially, if you upload any media to Google’s Search functionalities, it is utilized for AI training unless you choose to opt out.

The modification was introduced through a discreet update to Google’s Search services privacy settings, communicated in June via an email to users. With this update, the company effectively enrolled users in this broader AI training under the pretense of offering greater control over their saved history and tailored recommendations.

This update brought forth two additional settings, Search Services History and Personalized Recommendations, enabling you to adjust how your activity personalizes your Google experience and the duration your web and app activity is retained.

The update extends beyond Google Search itself, encompassing other search functions like Maps, Shopping, Flights, Hotels, Translate, and News.

For example, when you use Google Lens to conduct a visual search by taking a photo, that image might now be stored for AI training.

Likewise, if you utilize the newer Search Live feature for voice search in the Google app, those audio recordings might be kept, along with any other Google voice searches. If you engage with Google Translate to practice speaking, that audio is also recorded.

These adjustments signify a larger trend in the industry toward collecting data through various means to enhance AI services. Rather than relying solely on data gathered from the web, Google and similar companies are increasingly amassing information that people upload or generate during their service usage. Meta exemplifies this trend by leveraging users’ images and media, as well as content captured by its AI glasses, for AI training on a larger scale.

Google directly acknowledges the media training aspect, stating in the customer email: “Similar to your Search Services History, your saved media is utilized to develop and enhance Google services and technologies, which includes AI models and safety measures.”

The help documentation reiterates this, stating that the company “utilizes your history to provide, develop, and improve its services (such as training generative AI models) and to safeguard Google, its users, and the public with the assistance of human reviewers.”

Some of this data retention is temporary and linked to the operational functionality of the product, but according to Google’s own declarations, saved media can also be specifically held for the purpose of training its AI.

Modifying your settings

The positive aspect is that you possess some control over this. You can alter your preferences on the Search Services History and Search Services Personalization pages. For the former, you can uncheck the “Save Media” option independently from the “Search Services History” option, or opt out of both. You can also set how frequently you want your saved data to be automatically deleted — after three, 18, or 36 months.

Afterward, you can visit this page to explore additional privacy settings, including Web & App Activity, Timeline, YouTube History, and more.

In addition to saved media, Google also utilizes your search history, location, and various data from websites you browse to tailor your experience on Google, which includes the ads displayed.

Prior to this update, Google allowed you to manage how historical search data was retained via its “Web & App Activity” settings. That has now been divided into two distinct settings: the Web & App Activity data and the newly implemented Search data setting, which is enabled by default.

Consequently, if you alter the Web & App Activity data retention settings to opt out of your data being stored by the tech giant, the update will no longer affect your usage of Google Search services, as it is now a separate setting.

Providing a unique insight into the focus areas of a leading intelligence agency, Canada’s Communications Security Establishment revealed that it executed several state-sanctioned hacks last year aimed at disrupting the activities of drug dealers, violent extremists, and a ransomware group.

The revelations in the Canadian intelligence agency’s yearly report highlight some of the primary national security challenges confronting Canada and its allies: from the trafficking of illegal substances to cyber threats. The agency, CSE, is responsible for gathering foreign intelligence, safeguarding governmental systems, and countering online threats.

Released last week, the report indicates that the CSE performed three foreign “active cyber operations” last year — a term the agency employs to describe its cyber offensives against external threats to Canadian national security and public safety.

One operation mentioned in the report targeted cybercriminals outside Canada involved in the trade of chemicals necessary for producing the synthetic opioid, fentanyl. The CSE gathered intelligence on these brokers and subsequently executed an operation that “disrupted and reduced their operational capabilities,” as stated in the report.

Another active operation focused on gathering signals intelligence — information generated by electronics and devices connected to the internet — pertaining to a foreign extremist organization that was disseminating violent ideologies and enlisting members, including in Canada.

The report detailed how the agency assessed the group’s structure, influence, and potential weaknesses to carry out an operation that “effectively undermined the group’s credibility and restricted their capacity to radicalize and recruit new followers.”

Another initiative aimed at dismantling a ransomware-as-a-service operation that allowed hackers to lease access to a ransomware group’s infrastructure for launching damaging extortion schemes. The CSE noted that its signals intelligence unit uncovered the gang’s methods of targeting the healthcare, transportation, and business sectors in Canada, then employed an active cyber operation that “rendered the group’s infrastructure inoperable.” This operation also eliminated much of the information stored on the gang’s servers.

The agency reported that it carried out simultaneous “technical disruptions” against 10 of the most significant ransomware groups threatening Canada to “render parts of their infrastructure unusable.”

The report did not specify the locations of the hackers, extremists, or ransomware group, nor the specifics of the operations employed by the CSE against them. While it is common for intelligence agencies to launch cyberattacks against adversaries, such operations are rarely disclosed or detailed to safeguard the tactics and strategies utilized.

Cyber Command, based in Fort Meade, Maryland, which conducts cyber operations for the U.S. government, frequently carries out “hunt forward” missions that involve deploying cyber teams to partner nations to secure their networks and disrupt adversarial cyber campaigns. The frequency of U.S.-led hunt forward operations has increased from a handful in 2018 to over two dozen by 2025.

Additionally, Canada’s CSE stated that it executed one defensive cyber operation over the past year aimed at a phishing campaign targeting Canadian federal government entities and other critical systems. The agency reported it thwarted the group’s infrastructure and “diminished their capacity” to target Canadians.