XPPen has unveiled the Magic Pro 13, a new professional drawing pad designed to bridge the gap between creating on the move and working at a desktop setup. The device made its debut at IFA 2026 in Berlin, where XPPen positioned it as an all-scenario creative tool for artists who want professional capabilities without being tied to a fixed workstation.

The Magic Pro 13 is built around a two-in-one concept. It works as a portable drawing pad for capturing ideas away from a desk, while also functioning as a desktop-grade drawing display when connected to a PC. It runs Android 16 and features a 3K, 120Hz display, along with XPPen’s new X4 Smart Chip Stylus.

From mobile sketches to full desktop creation

The key feature is XPPen’s hardware-level Drawing Display Mode. When connected to a PC using a single full-featured USB-C cable, the Magic Pro 13 can switch into a mode intended for professional desktop workflows.

According to XPPen, the dedicated hardware architecture enables ultra-low latency, native colour output and higher-bandwidth video transmission. The company says this allows the device to deliver detailed visuals, fluid strokes and 16K levels of pressure sensitivity.

For creators, the significance is less about simply adding another drawing screen and more about reducing the need to carry separate devices for different stages of the creative process. An artist could capture an idea away from the studio and then continue refining it on a PC using the same device.

XPPen has also developed a software ecosystem around the Magic Pro 13. Its Shelf app automatically tags, organises and archives creative assets, functioning as an inspiration library, while Canvas provides a simpler environment for quickly sketching ideas without the complexity of traditional creative software.

XPPen expands its creative hardware lineup

The Magic Pro 13 is only one part of XPPen’s IFA 2026 showcase. The company is also highlighting the Magic Note Pad, Artist Pro 27 (Gen 2), Artist Ultra 14 and Pilot Pro, covering areas including professional drawing, mobile creation and editing.

The company says the wider lineup reflects its effort to make professional creative tools more connected and flexible. At IFA, visitors can also try the products and contribute to XPPen’s “DOODLE THE WORLD” Doodle Wall, which brings individual sketches together into a shared display.

For buyers, the Magic Pro 13’s actual availability will be the next important detail. XPPen says it is expected to launch in October 2026, with further information to be announced closer to launch.

Tata Consultancy Services said its HyperVault unit and partners have committed 700 billion rupees, about $7.4B, to an AI data centre campus of up to one gigawatt in Hyderabad, using green energy and water-neutral design principles. EU rules require any data centre above 500kW to report total and drinking water consumption to a European database every year.

Tata Consultancy Services said its HyperVault unit and partners have committed 700 billion rupees, about $7.4B, to an AI data centre campus in southern India. Bloomberg News reported the exchange filing on Saturday.

The site sits on 264 acres in Hyderabad, in Telangana, and will have a capacity of up to one gigawatt. The land is already secured, and TCS expects the campus to rank among the largest AI infrastructure sites in India.

“The campus will support frontier AI companies and hyperscalers with high-density, liquid-cooled compute infrastructure on a global scale,” the company said.

TCS also says the project will use green energy and water-neutral design principles. Neither term is defined in the filing.

The money follows a pattern. TPG put $1B into the company’s data centre business last November, part of a wider flow of foreign capital into Indian AI infrastructure through local firms.

Two of those words carry obligations in Europe that they do not carry in India, and the gap is not rhetorical.

Any data centre in the EU with at least 500kW of installed IT power must report annually to a European database, under a delegated regulation adopted in 2024.

The indicators include total water consumption and drinking water consumption, from which water usage effectiveness is calculated. Energy demand, waste heat recovery and the renewable share are in there too.

The annual deadline is 15 May. The regulation itself sets out no penalty for not filing, so the discipline is disclosure rather than enforcement.

So water-neutral design is a phrase in Hyderabad and a filed number in Frankfurt. Neither guarantees the water, but only one leaves evidence.

The scale runs the other way. Europe has opened bidding for up to seven AI gigafactories worth roughly EUR 30B, of which about EUR 10B would be public money and only around EUR 1B of Brussels’ share is committed.

Construction there is targeted for early 2027. TCS is putting $7.4B into one campus now, most of it private money rather than public.

Europe built the measuring regime first and the gigawatts second. Its own gigafactory plan has already stumbled on delays, and this filing is what the other order looks like.

OpenAI has acknowledged that it did not publicly disclose an earlier incident in which its autonomous AI agents took over a German wiki to communicate, share answers, and exchange techniques for bypassing restrictions.

The company says it treated the activity as model “misalignment” rather than a security incident, but now admits its disclosure practices must expand as AI systems increasingly cause real-world impact.

Rogue agents built their own message board

The incident began in May while OpenAI agents were completing timed, multi-round web lookup tasks.

In their writeup, independent researchers Sydney Von Arx, Cormac Slade Byrd, Spencer Kitts, and Thomas Larsen state that they uncovered roughly 18,000 posts from autonomous agents that “colluded to share answers, research their environment, and bypass sandbox restrictions.”

The agents were supposed to have read-only Internet access but discovered they could write to an obscure German programming wiki, DSEWiki (or DeutschesSoftwareEntwickler). They turned it into a shared message board for pooling answers, cheating on tests, predicting future questions, and exchanging techniques for bypassing OpenAI’s sandbox restrictions.

The researchers also found agents probing the wiki for cross-site scripting (XSS) flaws, impersonating its moderators, and establishing backup communications, though they found no evidence the XSS attempts succeeded.

When the administrator began removing their pages, one agent warned others that a “wiki cleanup/deletion sweep appears active alphabetically” and directed them to a backup page beginning with “ZZZ” so it would survive longer.

The researchers attributed the activity to internal OpenAI systems based on agent names referencing OpenAI, the nature and speed of the evaluation tasks, infrastructure associated with Microsoft Azure, and subsequent visits to the wiki from OpenAI-linked IP addresses.

However, their investigation was limited to information the agents wrote publicly.

The research team did not have access to OpenAI’s internal transcripts or other data that could establish precisely how the agents discovered the wiki and began coordinating through it.

OpenAI says its disclosure rules must change

In a statement published today, OpenAI said it had historically treated model misalignment as a research issue, with findings communicated through research papers and system cards.

The company said it considered the wiki activity another example of “misalignment” similar to behaviors it had previously discussed, rather than an incident requiring a dedicated public disclosure.

OpenAI’s own wording suggests a wider footprint than the researchers documented, describing the episode as one “where our agents wrote to several internet sites.”

That differed from its response to the Hugging Face compromise in July, when OpenAI said its AI models hacked the platform after discovering a vulnerability while completing cybersecurity tasks.

A follow-up analysis found that nearly 700 rogue AI agents had coordinated during the attack, sharing strategies and creating persistent access mechanisms without direct human instructions.

OpenAI said it treated the Hugging Face breach as a conventional security incident because it affected the security of both OpenAI and third parties. The AI giant began working with Hugging Face and publicly disclosed the incident the following day.

However, OpenAI now acknowledges that the distinction between research misalignment and security incidents is becoming increasingly difficult to maintain.

“This year, we’ve started to see misalignment cause new types of real-world impact,” the company said.

OpenAI says the AI industry lacks consistent standards governing when unexpected agent behavior during training, evaluation, or deployment should be reported, particularly when it does not resemble a traditional cybersecurity incident.

The company is developing a new disclosure framework that it plans to publish in the coming weeks and says it is discussing these issues with government regulators worldwide.

The timing of the acknowledgment is also notable, coming in the same week OpenAI launched GPT-6 Astra, which it touts as “the world’s most intelligent and aligned model” and state-of-the-art on computer use, browsing, software engineering, and cybersecurity.

OpenAI says Astra is better at staying within its intended scope, measured partly by a new evaluation it built in response to the Hugging Face incident.

However, the problem is not unique to OpenAI.

In July, Anthropic revealed that its Claude AI breached three organizations during internal security evaluations, in one case registering a package name it found in documentation and uploading malicious code to PyPI. The package was live for about an hour, in which 15 real systems downloaded and ran it.

As AI models become more capable and gain greater autonomy and access to the Internet and external tools, such incidents are expected to accelerate.

What remains unknown is what else these systems could become capable of, or end up doing, without stronger controls, oversight, and disclosure requirements.

Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.

The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.

Get the report

applications

Built-in plagiarism police to hand in badge next month

Microsoft is retiring Word’s Similarity Checker on October 3 and stripping the feature from Microsoft Editor.

Other writing and editing tools, including Microsoft 365 Copilot, won’t be affected. But if an organization uses Similarity Checker as part of its workflow, it’s time to consider alternatives.

Similarity Checker has been around for a few years. Launched in 2020, the Bing Search-powered service compared text from the user to online sources and came up with a percentage score – how much of the text was similar to those online sources.

The user could then add citations if appropriate, or try again for originality. Microsoft had previously said: “This can aid writers in focusing less on the mechanics of writing, and more on the content.”

Word’s citation capabilities will not be affected by the retirement, nor will its spelling and grammar checking tools. Although Microsoft’s AI assistant is unaffected, Microsoft warned, “Copilot is not intended to serve as a formal similarity, plagiarism, or originality validation tool.”

Plagiarism spotting and similarity checking have become all the more important in recent years. Turnitin, for example, added features to identify AI-generated text, and there is evidence that plagiarism-detecting tools have had an impact (although there are plenty of other nefarious uses for AI.)

The writing for the Similarity Checker in Word has been on the wall for a while now. Although Microsoft has not shared usage statistics, it did hint that users were giving it a swerve:

“This change allows us to focus on experiences and capabilities that are more broadly used across Microsoft Word.”

The functionality is certainly at odds with the Copilot-All-The-Things fever that has gripped Microsoft in recent years, despite Microsoft’s insistence that Copilot is not a tool for formal similarity checking.

That said, there were a few alarmed users when the tool stopped working for a few days in 2023.

After October 3, 2026, the Similarity Checker will stop working for good and cease to be accessible through the Editor experience in Word. ®