OpenAI Upgrades Codex With Computer Control, Image Generation, and Memory Features

/in /tarafından

OpenAI Codex Update Adds Computer Use, Image Generation, and Memory on Mac - MacRumors

OpenAI has announced a major upgrade to its Codex platform, expanding its capabilities far beyond traditional coding assistance. The update introduces a new generation of features designed to make the tool more versatile, interactive, and useful for developers working across different environments.

One of the most significant additions is “computer use,” which allows Codex to interact more directly with system-level tasks, alongside new image generation capabilities and native web access. These enhancements aim to transform Codex from a purely code-focused assistant into a broader productivity tool that can handle multiple types of digital workflows.

The rollout will begin on macOS first, with plans to extend support to Windows and integrated development environments (IDEs) in the near future. This phased approach ensures stability while gradually expanding access to a wider developer base. According to the company, the goal is to make Codex more seamlessly integrated into everyday development environments.

The timing of the update is also notable, arriving on the same day as the release of Anthropic Claude Opus 4.7, which also focuses on improvements in software engineering capabilities. This highlights the growing competition in the AI coding assistant space, where platforms are rapidly evolving to offer more advanced features.

OpenAI also revealed that Codex is now used by more than three million weekly active developers, underscoring its growing adoption. With this latest upgrade, the platform continues to position itself as a key tool for both individual developers and enterprise users seeking more powerful AI-assisted coding solutions.

Canva AI 2.0 Launches With Agent-Like Features and Design Memory Upgrade

/in /tarafından

Canva introduces Canva AI 2.0 with new conversational and agentic capabilities

The visual design platform Canva has officially unveiled a major upgrade to its artificial intelligence system with the launch of Canva AI 2.0. Announced at the Canva Create 2026 event in Hollywood Park, Los Angeles, the new version marks a significant step forward in how users interact with design tools through AI-powered assistance.

Unlike earlier versions, Canva AI 2.0 introduces a more advanced “agentic” system capable of understanding layered design elements and executing complex creative tasks through simple conversational prompts. This means users can now describe what they want, and the AI can handle multi-step design processes with far less manual input.

One of the standout additions is its ability to manage multi-channel campaign creation, allowing users to generate coordinated designs across different formats and platforms. The system also introduces memory capabilities, enabling it to remember user preferences and previous design choices to deliver more personalized results over time.

Beyond design generation, the upgraded suite includes features such as web research integration and task scheduling, expanding its role beyond a traditional creative tool into a more versatile productivity assistant. These enhancements are aimed at streamlining workflows for both individual creators and professional teams.

According to Canva, the rollout of Canva AI 2.0 will begin in phases over the coming weeks, eventually becoming available to users globally. The update reflects the company’s broader push toward making design more intuitive, automated, and accessible through artificial intelligence.

Apple Tap-to-Pay Security Flaw Shown in Video: Locked iPhone Used in $10K Demo

/in /tarafından

Video shows how to steal $10,000 from locked iPhone in controlled setting - 9to5Mac

A new security demonstration has raised concerns about Apple’s payment ecosystem after a YouTuber showcased a potential vulnerability affecting tap-to-pay functionality on iPhones. The issue reportedly allows a locked device to be manipulated in a way that simulates a legitimate payment interaction, without requiring the phone to be unlocked or directly authorized by the user.

Apple has long promoted its ecosystem as highly secure, particularly when it comes to features like NFC-based payments and Express Transit Mode. However, the demonstration suggests that under specific conditions, an attacker could potentially exploit a “man-in-the-middle” style setup to trick an iPhone into processing a payment as if it were interacting with a real point-of-sale terminal.

In the video, the creator—working alongside cybersecurity researchers—shows how specialized hardware can intercept and relay NFC signals. The setup reportedly uses a device connected to a computer to mimic a payment terminal, effectively convincing the iPhone that a legitimate transaction is taking place. The demonstration includes a test scenario involving a locked device, raising questions about how such interactions are validated.

The exploit, according to the video, is based on a concept that has been known in cybersecurity circles for several years, though it has now been presented in a more practical, real-world context. It highlights how NFC-based systems, while convenient, can still be vulnerable to relay attacks if security layers are bypassed or misinterpreted.

While there is no indication of widespread real-world abuse, the demonstration has sparked renewed discussion about mobile payment security and the importance of layered protections. It also underscores the ongoing challenge for companies like Apple to balance convenience features such as tap-to-pay with evolving security threats in mobile ecosystems.