PDF

Univer is an open-source TypeScript project from dream-num that bills itself as an 'Office Harness for AI Agents'. It provides a single runtime combining spreadsheets, documents, slides, canvas, relational tables, and PDF handling. The repository is trending on GitHub, and the framing suggests developers are interested in giving AI agents tools to create and manipulate office-style documents. Beyond the project's own description, there is little discussion in the available evidence explaining what users are saying about it.

A new cryptography paper describes a method for forging 1024-bit RSA signatures at a cost close to that of the Special Number Field Sieve, an algorithm for factoring large integers. If confirmed, the result would significantly lower the practical security margin of 1024-bit RSA keys, which are still found in older systems, and renew calls to migrate to 2048-bit keys or elliptic-curve cryptography.

A new research paper analyses the privacy practices of web and mobile conversational AI agents, examining what user data these tools collect and share. The paper, titled 'Prompt Like a Butterfly, Sting Like a Tracker', is drawing attention among developers and privacy-conscious readers who are weighing how much personal information popular chatbot services expose through tracking.

A group of Nobel Prize-winning economists has signed a public letter voicing support for the proposed tax on billionaires' wealth in California. The letter, hosted by economist Gabriel Zucman, a leading advocate of wealth taxation, argues the levy is economically sound. Supporters say endorsements from such prominent economists strengthen the measure's credibility against critics who claim it would drive the wealthy out of the state.

An open letter from Iran addressed to people in the United States and the European Union is circulating widely on social media. Shared on Mastodon, the letter—published as a PDF by Iran's official news agency IRNA—touches on themes of war, corruption, Palestine, Israel and history. Those sharing it describe it as extraordinary and are urging Western audiences to read it in full.

A 2014 paper recounts how Los Alamos National Laboratory turned to the ENIAC computer in 1947–1948 to run Monte Carlo simulations for nuclear weapons research. The work, tied to figures like Nicholas Metropolis and Stanislaw Ulam, helped establish Monte Carlo methods as a foundation of modern computational science. Readers are sharing and discussing the historical account of this early computing milestone.

The OpenID Foundation has released a 2025 white paper on identity management for agentic AI, addressing how autonomous AI agents should be authenticated, authorized and held accountable when acting on behalf of users. The paper, published as a PDF on the OpenID.net site, is drawing attention from developers and security professionals debating whether existing identity standards like OAuth can cope with machine-driven agents.

The latest Sum'Eau weekly monitoring bulletin on SARS-CoV-2 circulation in France has been released, covering wastewater data for week 38 of 2026, the week of 14 September. The report tracks virus levels in sewage across French cities, with results listed alphabetically and shared in PDF and HTML formats. Wastewater surveillance remains a key indicator of COVID-19 trends in the country.

An author is publishing his book 'Apocalypse No' chapter by chapter on social media, currently sharing Chapter 3, page 26, with a free PDF download available. The book weaves together climate concerns, ecopsychology and theories about human psychology, promoting themes of rejecting apocalyptic thinking. Engagement so far is modest, with small audiences following the serialisation across federated platforms.

The OpenID Foundation has released a 2025 white paper on identity management for agentic AI, addressing how autonomous software agents can be authenticated and authorized when acting on behalf of users. The paper examines the gaps existing identity standards like OAuth and OpenID Connect leave when AI agents, rather than people, initiate actions across services. Developers and security professionals are weighing how current identity infrastructure can adapt to machine-driven agents.

The OpenID Foundation has released a paper on identity management for agentic AI, addressing how autonomous AI agents should authenticate and be authorized when acting on behalf of users. The document, published in October 2025, is circulating among developers and security practitioners who are debating how existing standards like OpenID Connect might extend to machine actors.

A software discussion is making the case that true redaction of PDF invoices must be treated as a destructive transformation: remove the sensitive text first, then apply the digital signature to the altered document. Drawing black boxes over content, by contrast, leaves the underlying data intact and recoverable. The point matters in fintech and banking contexts, where invoices are shared widely and hidden text can leak confidential information despite appearing covered.

A critical remote code execution vulnerability in Stirling PDF, tracked as CVE-2026-85714 and rated 9.1, has full technical details and a proof-of-concept exploit publicly available. The flaw stems from a crafted SQL import affecting the bundled H2 database. Administrators are urged to upgrade to version 2.13.2 immediately, as the public exploit makes attacks likely.

The OpenID Foundation has released a paper on identity management for agentic AI, addressing how autonomous AI agents should authenticate and be authorized when acting on behalf of users. The document examines how existing standards like OAuth and OpenID Connect can extend to machine agents that independently make decisions and take actions, a growing concern as companies deploy AI agents with access to sensitive systems and credentials.

A newly disclosed security flaw affects Apple devices: opening a malicious PDF file can cause an iPhone, iPad or Mac to crash. Security outlet The Hacker News reported the issue, and users are being advised to install the latest software updates to keep their devices protected until a full fix is in place.

A guide circulating among security-minded users explains how PDF redaction can fail: drawing a dark box over a line hides it visually, but the underlying selectable text remains in the file. It recommends preparing a separate sharing copy and exporting redacted pages as new images, so unapproved text is genuinely removed rather than merely covered, before documents are released.

OpenAI has rolled out new capabilities for ChatGPT, including an AI-powered virtual try-on feature that lets users preview how clothing or products look on them, and a camera-based function that scans documents and converts them into PDF files. The update, reported by ITmedia News, is being shared among Japanese tech audiences, with discussion framing it as another sign of OpenAI expanding ChatGPT from a chatbot into a broader everyday utility and agent-style assistant.

A newly circulated paper recounts how Los Alamos scientists turned to the Eniac in 1947–1948 to run some of the earliest Monte Carlo simulations, using the pioneering computer to model nuclear processes through random sampling. The episode is widely seen as a founding moment of computational science, showing how wartime computing hardware was repurposed for peacetime physics research and helped establish a technique now used across finance, engineering and beyond.

A German-language post circulating on Mastodon shares a portrait of Dr. Arthur Janov along with a PDF compiling summaries of article series from Janov's website, hosted on the site dieontogenetischeseite.de. Janov, who died in 2017, was the American psychotherapist known for developing primal therapy and the influential 1970 book 'The Primal Scream'.

A first public proof-of-concept is now available for CVE-2026-86950, a vulnerability in Apple's CoreGraphics. A malicious PDF with a crafted embedded font crashes unpatched iPhones and Macs, though no working exploit has been demonstrated. Apple patched the flaw on September 28 after targeted attacks, and CISA set an October 2 fix deadline for federal agencies under its Known Exploited Vulnerabilities catalog.

The FIA has published a new car display procedure for Formula 1, dated 1 October 2026, releasing the document as a PDF. The document is tagged with Formula 1 and FIA labels, indicating it is an official sporting or technical procedure. Fans and motorsport followers are sharing the release, with little commentary available so far on its contents or implications.

Developers are sharing methods to split large PDF documents using Python, a common task in offices dealing with consolidated quarterly reports, scanned contracts and other bulky files. The discussion highlights lightweight scripting approaches as a productivity boost for everyday document handling, and it is drawing interest from programmers and non-programmers alike looking to automate tedious work.

A technologist is promoting a small collection of self-published digital guides on data science and AI. The PDFs cover prompting workflows, data science resources and machine learning best practices, and are pitched as quick, structured reads for people who prefer guides over scattered blog posts or long textbooks. Response so far is limited, with only a couple of likes, suggesting it is early days for the effort.