Tech Showdown: Musk vs OpenAI, Military Smart Glasses, and AI's Next Frontier

Welcome to our deep dive into the week's biggest tech stories. We'll explore why Elon Musk's lawsuit against OpenAI failed, how Anduril and Meta are turning smart glasses into battlefield tools, what to watch for at Google I/O, and the rise of world models—a new AI paradigm that could reshape how machines understand reality. Each section answers key questions, with all facts drawn from the original report.

Why did Elon Musk lose his lawsuit against OpenAI?

A jury ruled in favor of OpenAI, deciding that Musk brought his case too late. The suit claimed OpenAI had abandoned its nonprofit mission by shifting to a for-profit structure, breaching its founding contract. However, the court did not examine whether OpenAI actually violated that mission. Instead, it focused solely on the timing of Musk's filing. The company argued that signs of a for-profit pivot were visible as early as 2017, while Musk said he only learned of the change in 2022. The statute of limitations had expired, so the case was dismissed. Legal experts say the verdict does not settle the broader debate about OpenAI's governance—it only closes this particular legal window. The fight over OpenAI's structure may resurface in other forums or future litigation.

Tech Showdown: Musk vs OpenAI, Military Smart Glasses, and AI's Next Frontier — Source: www.technologyreview.com

What was the central timeline dispute in the case?

The crux of the argument revolved around when OpenAI began shifting from its original nonprofit model toward a for-profit entity. OpenAI presented evidence that internal discussions and structural changes started as early as 2017, with key decisions made public soon after. Musk countered that he was unaware of any shift until 2022, claiming he had been misled about the company's direction. The jury found that even if Musk's version were true, he should have discovered the alleged breach earlier through reasonable diligence. Because he filed in 2024, the statute of limitations had run out. This timeline battle meant the jury never weighed in on the ethics of OpenAI's transformation—only on whether the case was timely.

How are Anduril and Meta developing smart glasses for military use?

Defense contractor Anduril and tech giant Meta are prototyping an augmented-reality headset designed for soldiers. Quay Barnett, a former Army Special Operations officer leading the project at Anduril, explains that the goal is to optimize “the human as a weapons system.” The headset will incorporate eye-tracking and voice commands, allowing users to perform actions such as ordering a drone strike simply by looking at a target and speaking a command. Meta brings its expertise in consumer AR hardware, while Anduril integrates battlefield systems. The device is still in prototype phase, but the vision includes real-time data overlay, enhanced situational awareness, and seamless coordination with unmanned systems. Critics raise concerns about automating lethal decisions and the escalation of warfare pace.

What capabilities will these military smart glasses have?

The planned smart glasses will allow soldiers to control drones and other equipment using eye movements and voice. For instance, a user could look at a target and say “engage” to authorize a strike. The headset will overlay tactical information—maps, enemy positions, unit locations—directly onto the wearer's field of view. It will also support hands-free communication, reducing the need for bulky radios and screens. The system aims to accelerate decision-making in combat, but it also raises ethical questions about delegating lethal force to human-machine interfaces. Anduril has not disclosed a timeline for deployment, but testing is underway with select military units. The partnership with Meta suggests the device could eventually leverage consumer AR advances, though military hardening will be required for field use.

What can we expect at Google I/O this year?

Google's annual developer conference arrives with the company facing intense competition in AI. Many analysts rank Google a clear third in foundation models, behind OpenAI and Anthropic. Coding tools like Claude Code and OpenAI's Codex have outpaced Google's offerings. To counter this, Google is expected to announce major updates to its Gemini model, improved coding assistants, and deeper integration of AI into its products. The company also leads in AI for science—areas like protein folding and materials discovery—and will likely showcase new research. Executive speeches will emphasize a dual strategy: catching up in coding and pushing boundaries in scientific AI. Hardware announcements, such as a new Pixel device or Android updates, are also possible. The event runs for two days, with keynotes and breakout sessions.

Why is Google considered third place in foundation models?

Foundation models today are largely judged by their coding abilities. OpenAI's Codex and Anthropic's Claude Code set the standard for generating and debugging code. Google's Gemini models, while strong in language understanding and science, have lagged in coding benchmarks. Industry observers note that developers prefer Claude and GPT-4 for programming tasks, citing higher accuracy and better tool integration. Google's reputation in AI research remains stellar, but market perception has shifted. The company still shapes cutting-edge areas like multimodal models and AI safety, but it has lost the public narrative war. At I/O, Google will attempt to close the gap by demonstrating rapid improvements in coding performance and launching new developer tools that rival its competitors.

What are world models and why are they gaining attention?

World models are a new class of AI systems designed to understand and simulate physical environments, unlike large language models (LLMs) that only process text. They learn how objects interact, how scenes change over time, and what actions are likely to produce certain outcomes. Recent advances from Google DeepMind, Fei-Fei Li's World Labs, and Yann LeCun's startup have made world models a hot topic. These systems could enable robots to navigate unfamiliar spaces, power immersive simulations, and help AI reason about cause and effect. Because they model the world rather than just predicting words, they promise more robust and general intelligence. However, they are still early-stage and computationally expensive. Researchers believe they may overcome some of LLMs' limitations, such as hallucinations and lack of common sense.

What upcoming event will cover world models?

MIT Technology Review is hosting a virtual event on Thursday, May 21, to examine world models. Editor in chief Mat Honan will lead a discussion featuring senior writers and researchers. The event will explore progress from leading labs, the potential for robotics and simulation, and the challenges ahead. Attendees can expect a balanced look at hype versus reality. Registration is open to subscribers. This event comes at a crucial time as world models move from academic papers to commercial experiments. It will also address how these systems might integrate with existing AI stacks and what they mean for the broader AI race.