NerdNews

OpenAI and Apple's partnership has become strained, with OpenAI looking to blame Apple for its lack of integration with iOS. The company expected ChatGPT to be more deeply ingrained in the ecosystem, but Apple has had trouble making its AI features work. Apple plans to open up its AI relationships, potentially leaving OpenAI as just one of many AI models available on the platform.

The trial between Elon Musk and OpenAI has concluded with closing arguments. The case revolves around the accusation that OpenAI abandoned its founding mission in pursuit of profits. While the jury's decision is pending, the trial has revealed damaging information about the characters of key players, including Elon Musk and OpenAI's CEO Sam Altman.

Recursive Superintelligence, a San Francisco-based startup, aims to create a recursively self-improving AI model that can autonomously identify its weaknesses and redesign itself. The startup is founded by Richard Socher and has raised $650 million in funding. The goal is to achieve recursive self-improvement, where the AI system can improve itself without human involvement.

Cerebras Systems raised $5.5 billion in its IPO, with shares priced at $185 and opening at $385, more than double the IPO price. The company's valuation reached $66 billion, with its founders' stakes worth nearly $1.9 billion and $1 billion, respectively. Cerebras is a major contender for supplying chips for AI inference and counts OpenAI, G42, and Amazon Web Services as customers.

A zero-day exploit called YellowKey allows attackers to bypass default Windows 11 BitLocker protections, giving them full access to an encrypted drive. The exploit works by using a custom-made FsTx folder on a USB drive to manipulate the Windows Recovery environment, allowing an attacker to access the drive without a BitLocker recovery key.

This article explores the fundamentals of time-series storage design, including normalization, high cardinality, schema evolution, columnar storage, and partitioning. It provides practical guidance on optimizing storage footprint, query performance, and cost, and discusses the trade-offs between different design choices.

Anthropic's Bun, a JavaScript toolkit and runtime, has been rewritten in Rust from Zig. The merge happened quickly, with over 1 million lines of code added, and is expected to improve performance and reduce memory leaks. The rewrite was accomplished using AI tools, which has raised questions about the role of AI in software development.

A survey by Harness found that 89% of developers believe productivity metrics have improved with AI, but 81% spend more time reviewing AI-generated code, increasing friction and decreasing productivity gains.

OpenAI has released its Codex coding app on mobile, allowing users to access their coding projects through the ChatGPT app on Android and iOS. The app acts as an intermediary between the user's device and their coding environment, keeping files and credentials secure. Users can prompt Codex from their phone and receive updates, including screenshots and test results, reducing downtime on projects.

OpenAI has integrated its Codex desktop AI tool into the ChatGPT mobile app, allowing users to access and control Codex from their iOS or Android devices. This feature enables users to work on tasks, review outputs, and approve commands remotely. The integration is currently available as a preview for all ChatGPT plans, including the free plan.

OpenAI's coding tool Codex is being integrated into the ChatGPT app, allowing users to monitor and manage development workflows remotely from their phones. The update is currently in preview and available on iOS and Android.

Cerebras Systems, a company that makes wafer-scale AI accelerators, has gone public with a successful IPO, raising $5.55 billion and valuing the company at $66 billion. The company's unique approach to chipmaking, using a wafer-scale design, has allowed it to deliver high-performance AI computing solutions. Cerebras' technology has been adopted by several major companies, including AWS and OpenAI.

Google has launched a new feature called Android Intrusion Logging, which helps investigate spyware attacks on Android devices. This tool is part of Google's Android Advanced Protection Mode and allows high-risk users to log device and network activities for forensic analysis. The logs include security events, spyware installation, and DNS connections, and are encrypted and securely archived in the user's Google account.

Microsoft is trying to stay out of the drama in the Musk v. Altman trial, with their lawyers frequently stating that the company was not involved in key decisions. Microsoft was an early funder of OpenAI's for-profit company, but they have maintained a low profile throughout the trial. The company's CEO, Satya Nadella, testified that OpenAI's board drama was 'amateur city' and Microsoft has largely avoided getting entangled in the controversy.

Anthropic's Mythos AI model has been able to hack macOS, bypassing Apple's security systems by linking two bugs together to corrupt the Mac's memory. This allowed the AI to gain access to parts of the device that should be inaccessible. The exploit was achieved with the help of human researchers and targeted Apple's Memory Integrity Enforcement system.

OpenAI confirmed that hackers stole some data after a security issue with its code. The company said two employees' devices were impacted, but found no evidence of compromise to user data, production systems, or intellectual property. The attack is part of a string of recent supply-chain attacks targeting software developers and their projects.

Wirestock, a platform that supplies datasets of images, videos, design assets, and gaming and 3D content to AI labs, has raised $23 million in Series A funding. The company has pivoted from a creative marketplace to a data provider, and currently provides multimodal data to six of the largest foundation model makers. Wirestock has signed up over 700,000 artists and designers who complete tasks for data collection, and has paid out $15 million to its contributors.

The Musk v. Altman trial is deciding whether Elon Musk will win his lawsuit against OpenAI and Sam Altman. The trial has made both parties look bad, with evidence suggesting that the nonprofit structure was used to attract talent and garner goodwill, but was seen as a roadblock to building a massive business. The outcome of the trial could have significant implications for the future of AI development.

OpenAI is exploring legal action against Apple due to disappointing financial results from ChatGPT integrations, which didn't drive expected subscription growth. The partnership allowed ChatGPT to be used within Apple's ecosystem, but the implementation was more limited than OpenAI had hoped, with fewer features and tighter controls than the standalone app.

An 18-year-old flaw in NGINX can be exploited for denial of service and potential remote code execution. The vulnerability, tracked as CVE-2026-42945, has a critical severity rating of 9.2. It affects NGINX versions 0.6.27 through 1.30.0 and can be triggered when NGINX configurations use both the ‘rewrite’ and ‘set’ directives.