📰 English IT Daily · 2026-06-23

A new article by digital artist David Revoy explains why some drawing tablet brands are not cooperating on free Linux drivers. Revoy often reviews tablets on GNU/Linux and uses only free and open source software. During these tests, he collects hardware information and shares it with developers who work on Linux input support. Their goal is to build high-quality drivers, which are programs that let the operating system communicate correctly with hardware devices.

Revoy says this process takes a lot of time. He has to test the tablet, collect technical data, check how the driver works, and then prepare a video review and a blog post. Because of this workload, he tried a different strategy. Instead of acting as a middle person, he wanted tablet brands to work directly with Linux developers and share device specifications. He pointed to Wacom as an example of a company that has supported Linux in this way for many years.

After many emails, Revoy was finally connected to a technical contact through Gaomon. He says this contact appeared to be linked to Huion, which was interesting because he had already noticed similarities in the proprietary Linux driver packages used by several brands, including Gaomon, XpPen, Huion, and Ugee. This made him think one technical team might be responsible for drivers across multiple brands. He shared information about the open project and encouraged contact with experienced Linux input developers.

However, the result was a polite rejection. According to the message he received, the company decided not to join the Linux driver project at this time. The reasons included limited business impact and concern about sharing device specifications with a project they viewed as being led by a competitor, Wacom. The case shows a common problem in tech: even when open collaboration could improve support for users, companies may avoid it because of branding, competition, and control over proprietary technology.

A new debate is growing around the cost of artificial intelligence services. Some analysts say AI platforms became popular partly because they offered very low prices compared with the real cost of running the systems. This strategy can quickly attract users and increase demand. However, critics argue that it may also hide a serious business problem if companies cannot earn enough money to cover their spending.

The concern is simple: advanced AI requires expensive chips, data centers, electricity, and engineering work. Reports discussed in the source article suggest that some subscription plans may let customers use far more computing power than the monthly fee seems to support. In other words, companies may be subsidizing usage by charging less than the true cost. A subsidy is financial support that lowers the price for the customer. This can help growth in the short term, but it may create pressure later.

The source also refers to outside analysis comparing token usage with subscription prices. Tokens are small units of text that AI models read and generate. If a customer can consume a very large number of tokens for a fixed fee, the platform may face weak gross margins, meaning little money is left after direct operating costs are paid. Recent financial reporting mentioned in the source added to these concerns by showing very high costs, large losses, and a big gap between revenue and spending.

For businesses, the affordability question matters because AI is moving from experiments to daily work. If prices rise in the future, companies that built workflows around heavy AI use could face difficult budget decisions. The issue is not whether AI is useful, but whether current pricing is sustainable. As more firms adopt AI for coding, search, writing, and support, they may need to watch usage carefully and ask whether today's low prices will still exist tomorrow.

A recent technical article explains that the enemy AI in Elden Ring may be less complex than many players expect. FromSoftware is famous for difficult and varied fights, but the code behind them appears to use a fairly simple design. According to the article, much of the behavior is written in Havok Script, a Lua-based scripting system for games. Because this code has been extracted and studied by others, it is possible to see the general structure of how the AI makes decisions.

The main idea is something called a goal. A goal is a state or task that an actor, such as an enemy or boss, can perform. Instead of using only a normal finite state machine, the system keeps a stack of goals. In simple terms, the AI works on the goal at the top of the stack. That goal can then add sub-goals, which are smaller actions to be done next. This makes the system more flexible than a basic state machine because it can handle short action sequences more naturally.

Each frame, the actor updates the current goal. The goal then returns a result such as continue, success, or failure. If it returns success, that goal is removed from the stack. If it returns failure, the system can remove not only that goal but also related unfinished sub-goals until it returns to the parent goal. For example, a boss may begin a combo attack as several connected sub-goals. If one attack fails, the AI can stop the rest of that sequence and choose a different action.

The article says that much of the real decision-making happens in an activate function. This function runs when a goal starts and again when its sub-goals have finished. At that point, the AI chooses the next action by looking at the game world, the actor’s condition, and some randomness. The result is an important lesson for software engineers: impressive behavior does not always require high-tech methods. A well-designed system with clear rules, reusable states, and limited randomness can still create rich and believable results.

Mozilla says many internet users are facing more CAPTCHAs, login requests, and blocked pages than before. The company argues that this is happening because websites are dealing with growing amounts of harmful bot traffic. Bots are automated programs, and some of them are used for attacks such as spam or credential stuffing, which means trying many stolen usernames and passwords at high speed. In response, websites are using stronger defenses, but these tools can also make normal users feel punished.

According to Mozilla, privacy improvements in browsers have created a difficult situation. Many browsers now reduce tracking by blocking third-party cookies, limiting fingerprinting, and hiding IP addresses. These steps help protect people from being followed online. However, the same signals were often used by anti-abuse systems to decide whether a visitor looked suspicious. As a result, some websites now ask more people to prove they are human, sign in, or stop using privacy tools if they want access.

Mozilla warns against solutions that require users to prove that their devices and software are officially trusted. It points to proposals such as Web Environment Integrity, saying they could shift power away from users and toward a small group of operating system and hardware companies. In Mozilla’s view, that would hurt user choice and weaken the open web, where people should be able to access sites without needing permission from a few gatekeepers.

Instead, Mozilla says the main problem with bots is scale, not identity. A website may not need to know exactly who a visitor is; it may only need to know that the visitor is staying within a reasonable rate limit, or usage cap. Mozilla is starting a new initiative with Cloudflare, other browsers, and web stakeholders to explore this idea while keeping the web anonymous by default. The goal is to reduce abuse without depending on identifiers that can also be used for tracking.

OpenAI has introduced DayBreak, a project focused on securing the digital world, and it also presented GPT-5.5-Cyber as part of that effort. The announcement shows how AI is being used not only for writing and coding, but also for cybersecurity. Cybersecurity means protecting computer systems, networks, and data from attacks. As digital threats grow more complex, many organizations are looking for tools that can help experts work faster and more accurately.

A cyber-focused AI model can support security teams in several ways. It may help analyze logs, summarize reports, explain technical findings in clear language, and identify possible risks in software or network activity. In simple terms, logs are records of what happened in a system, and risks are situations that could lead to harm. These tasks often take time, so automation can reduce routine work and let human specialists focus on difficult decisions.

At the same time, OpenAI’s message around DayBreak is also about responsibility. Powerful security tools can be useful for defenders, but they must be designed carefully to avoid misuse. In cybersecurity, misuse can mean helping attackers discover weaknesses or create harmful actions more easily. For that reason, companies usually add safeguards, testing, and human review before releasing advanced systems. This balanced approach is important because security technology can protect people, but it can also create new concerns if controls are weak.

The broader significance of DayBreak is that AI is becoming a more active part of modern defense work. Businesses and public institutions need better ways to respond to threats that move quickly across connected systems. An AI system like GPT-5.5-Cyber may not replace security professionals, but it could become a useful assistant for analysis, monitoring, and communication. For engineers, this trend raises an important question: how can organizations use AI to strengthen defense while keeping trust, safety, and oversight at the center?

Rhombus version 1.0 is now available, according to the Racket blog. The new language is built on Racket, a programming system known for strong support for language design and macros. Rhombus aims to combine that power with a more conventional syntax, so it can feel easier to read for people who do not want a parenthesis-heavy style. Developers can install it from the Rhombus website and learn through its documentation and example programs.

The Rhombus team says a new language is useful because modern software needs both common features and domain-specific tools. Many languages already offer variables, closures, objects, and pattern matching. However, not every language can be easily extended for special tasks. Rhombus focuses on extensibility, which means programmers can adapt the language with new constructs. In simple terms, a construct is a language feature or pattern that helps express an idea in code.

According to the project FAQ, Rhombus is a general-purpose, functional, extensible language with practical libraries and good performance. It is mainly a dynamic language, meaning types are often checked while the program runs, which can support interactivity and flexibility. At the same time, it also includes static and abstraction-enforcing tools that help code grow from small scripts to larger systems. The project also highlights functional data structures and compact repetition with ellipses as notable features.

Rhombus is closely related to Racket. The blog explains that using "#lang rhombus" in a Racket module makes it a Rhombus module, showing how it fits into Racket’s multi-language ecosystem. The relationship is compared to pairs like Elixir and Erlang or Kotlin and Java. For learners and engineers, Rhombus is interesting because it tries to make advanced language extension more accessible without giving up a familiar syntax. That could matter for teams that want both everyday productivity and deeper customization.

Oak is a new version control system created for software agents. A version control system tracks changes in code and helps teams manage project history. The developer behind Oak says current tools can slow agents down when they work on large or complex projects. Oak aims to improve both speed and context, meaning the information an agent needs to understand a task and continue work correctly.

One of Oak’s main ideas is virtual mounts. This means agents working locally or in the cloud do not need a full copy of a repository before they start. A repository is the stored collection of project files and history. In theory, this could let many tasks run in parallel more easily, because agents would not need to download everything or manage multiple worktrees just to switch between jobs.

The project is still early in development, so it is not yet a full replacement for older tools. According to the developer, Oak currently has no Windows build and is missing several common platform features, including CI, issues, and comments. CI, or continuous integration, is the practice of automatically testing and building code after changes. These missing parts show that Oak is still experimental, even if its design is ambitious.

Even so, the project has attracted attention because it reflects a larger change in programming. More developers are asking whether tools built for humans are also right for AI agents. Oak’s developer says the team has already been using Oak to build Oak for several months without a Git backup, while still relying on GitHub Actions for builds. If agent-based development grows, tools like Oak may push the industry to rethink how version control should work.

Mitchell Hashimoto has announced that his family will pledge another $400,000 to the Zig Software Foundation. With this new commitment, their total pledged support reaches $700,000, following an earlier donation in 2024. He said Zig continues to earn his respect both as a technical project and as a community. He also pointed to recent development updates as evidence of steady progress on the difficult work of building a programming language and compiler.

In his post, Hashimoto praised Zig’s approach to maintainership and community. He mentioned ideas such as Contributor Poker, an initiative connected to community participation, and Zig’s AI ban. A ban is a rule that does not allow something. In this case, Zig has a strict no-LLM contribution policy, meaning code or contributions created with large language models are not accepted. Hashimoto said this philosophy helps attract and develop talented people in open source.

The post also referred to a recent public debate about Zig’s policy, especially in relation to Bun’s Zig fork and Rust rewrite. A fork is a separate version of a software project that starts from existing code and then develops independently. Hashimoto said he respects Bun and did not want to turn the discussion into criticism of that project. Instead, he said he was struck by how quickly people treated each other as enemies online and how often empathy was missing from the conversation.

Hashimoto made it clear that he uses AI heavily and has written before about using it to ship real features. At the same time, he has also expressed concern about AI’s negative effects on open source. Although his views do not fully match the Zig Software Foundation’s policies, he said he still respects the people, the project, and its unusual boundaries. He added that Zig made it possible for him to build the kind of software he wanted, including Ghostty, and said that is why he continues to support the foundation.

A Danish privacy activist, Lars Andersen, said on social media that police raided his home. The post quickly drew attention from people interested in digital rights, privacy, and state power. Because the original public information is limited, many details about the case remain unclear. However, the report has already started a wider discussion about how governments investigate people who work on privacy-related issues.

Privacy activists often speak about surveillance, data collection, and the need to protect personal information. Surveillance means close monitoring of people, often using digital tools, cameras, or communication records. In democratic countries, police may search a home if they believe it is necessary for an investigation. At the same time, such actions can raise concern if the target is known for criticizing government systems or security practices.

This case also highlights the tension between public safety and civil liberties. Civil liberties are basic rights such as freedom of speech, privacy, and protection from unfair state action. When police use legal authority to seize devices or access digital records, experts usually ask whether the action was proportionate and transparent. In technology-related cases, the handling of encrypted data, online accounts, and security tools can become especially sensitive.

For people working in security and cloud systems, the story is a reminder that technology is never only technical. Engineers build systems that store logs, identity data, and user content, and these systems may later become part of legal requests or investigations. That is why clear access control, strong encryption, and careful data retention policies matter. Even when facts are still developing, cases like this encourage deeper debate about trust, accountability, and the balance between privacy and law enforcement.

The U.S. National Science Foundation, or NSF, is reportedly cutting or shrinking some research programs to make room for a new technology initiative. According to insiders cited by Science, the change could move money away from long-running basic research areas and toward projects linked more directly to national technology goals. Basic research means early-stage scientific work that does not always lead to a product right away.

Supporters of the shift may argue that the United States needs faster progress in important fields such as advanced computing, artificial intelligence, and other strategic technologies. In that view, federal funding should focus more on areas that can strengthen innovation, industry, and national competitiveness. A technology initiative is a planned effort to support a set of priority technologies through funding and policy.

However, many researchers worry that slashing existing programs could damage the broader scientific ecosystem. They say discovery often comes from curiosity-driven work, where scientists explore questions without knowing the final use. If funding becomes too narrow, universities may find it harder to support young researchers, train graduate students, and build the knowledge base that later supports major breakthroughs in engineering and science.

The reported NSF changes also raise a wider policy question: how should public agencies balance immediate national needs with long-term scientific capacity? For people in technology, the story is a reminder that innovation depends not only on products and startups, but also on stable research funding, clear priorities, and good governance. The final impact will depend on how deeply programs are cut and whether the new initiative creates value without weakening the foundation of future research.