WASHINGTON — “There’s this dark period between now and some time in the future where the advantage is very much offensive AI,” warns Rob Joyce, former director of cybersecurity at the National Security Agency. His observation, shared with Bloomberg, highlights a precarious moment in the evolution of Artificial Intelligence and global security.

As AI begins to actively shape kinetic conflicts, particularly in the tensions surrounding Iran, the gap between technological capability and human control is widening. The Pentagon currently maintains guidelines to keep “humans in the loop” to ensure security and accountability, but experts suggest this is a comforting fiction.

The true peril is not a rogue machine acting alone, but a human overseer who is fundamentally incapable of understanding the AI’s “thought process.” This systemic blindness creates a dangerous illusion of control, necessitating an urgent overhaul of safeguards around AI warfare.

The Power Struggle: Silicon Valley vs. The Pentagon

The friction between AI innovators and government regulators has reached a fever pitch. Trump administration officials are currently negotiating access to “Mythos,” a powerful new model from Anthropic, despite the company previously being blacklisted. Axios reports that the White House is eager for the model’s capabilities, even as Anthropic warns that Mythos was too dangerous for public release.

This tension is not merely technical but cultural. The Pentagon has reportedly engaged in a “culture war” against Anthropic—a strategy that many argue has backfired. While the BBC notes that finance ministers remain alarmed by security risks, Anthropic has attempted to pivot by releasing a less volatile model, as detailed by CNBC.

Parallel to this, OpenAI is facing internal and external scrutiny. Sam Altman’s opaque external investments have sparked conflict-of-interest concerns, with the Wall Street Journal reporting that these side hustles could compromise decision-making. Meanwhile, a legal battle is brewing over whether OpenAI has abandoned its original non-profit mission, even as the firm doubles down on its ambitions in the scientific community.

If the architects of these systems are fighting over control and ethics, what happens when the systems themselves become the primary tools of statecraft? Could we be delegating the fate of nations to algorithms whose logic we cannot decipher?

The Infrastructure of Intelligence

The expansion of AI is hitting a physical wall. The Financial Times reports that 40% of data center projects are currently delayed, partly due to local opposition—the classic “Not In My Backyard” (NIMBY) syndrome. This is a critical bottleneck, as modern AI’s appetite for power and space is insatiable.

Furthermore, the Pentagon’s reliance on private infrastructure has revealed glaring vulnerabilities. A recent Starlink outage during drone tests highlighted a dangerous dependency on SpaceX, according to Reuters. To diversify, the Department of Defense is looking toward legacy industrial giants like Ford and GM for military innovation.

On the global stage, the race for “world models”—AI that understands physical reality—is intensifying. Alibaba’s “Happy Oyster” model represents a significant leap in AI’s spatial comprehension, as noted by SCMP, although the Financial Times suggests that mastering cause-and-effect remains a hurdle.

The Consumer Shift: From Coding to Culture

AI is rapidly migrating from the lab to the living room. Google is streamlining the user experience by reducing the need for complex prompts through personal intelligence. In the developer world, OpenAI’s latest Codex updates are a direct challenge to Claude Code, as The Verge reports. However, the long-term viability of AI-driven coding remains a point of contention among senior engineers.

While some find hope in technology—such as AI-powered smartglasses bringing Korean theater to a global audience—others are fighting for their livelihoods. Global voice actors are currently resisting Hollywood’s AI push, arguing that their own voices are being used to train the very models intended to replace them, according to Rest of World.

Regulation is attempting to keep pace. Europe has launched a free age-verification app for companies, attempting to bring order to the digital wild west.

As these systems evolve, we must ask: are we creating tools that augment human potential, or are we architecting our own obsolescence?

<p>However, recent research by French geneticists suggests this may be a misunderstanding. They propose that these genetic markers are not the result of interbreeding, but of "population structure"—the natural concentration of genes within isolated groups. For more on how this shifts our understanding of <a href="https://www.technologyreview.com/2026/04/14/1135169/problem-thinking-part-neanderthal-human-evolution/?utm_source=the_download&amp;utm_medium=email&amp;utm_campaign=the_download.unpaid.engagement&amp;utm_term=*%7CSUBCLASS%7C*&amp;utm_content=*%7CDATE:m-d-Y%7C*">human evolution</a>, consider the broader implications of how we define "species."</p>

<p>This mirrors the current debate in AI: the difference between a system that *simulates* a pattern (population structure) and one that *incorporates* new information (interbreeding). Just as we are questioning our biological heritage, we are questioning whether LLMs are truly "learning" or simply reflecting the statistical structure of their training data.</p>

<p>This intellectual race is underpinned by a physical one: the quest for rare earth elements. The global transition to green energy and advanced computing depends on minerals that <a href="https://www.technologyreview.com/2024/01/05/1084791/rare-earth-materials-clean-energy/?utm_source=the_download&amp;utm_medium=email&amp;utm_campaign=the_download.unpaid.engagement&amp;utm_term=*%7CSUBCLASS%7C*&amp;utm_content=*%7CDATE:m-d-Y%7C*">critically power clean energy</a>. With China dominating the supply chain, the US and its allies are exploring unconventional extraction methods to avoid a strategic chokehold. For more on global resource security, the <a href="https://www.cfr.org">Council on Foreign Relations</a> provides essential context on the geopolitical implications of mineral dependency.</p>

<p>The evolution of intelligence—whether biological or silicon—is never a straight line. It is a series of disruptions, breakthroughs, and revisions. By studying the <a href="https://www.nature.com">Nature</a> of these shifts, we can better prepare for a world where the line between "human" and "machine" continues to blur.</p>

In the midst of this geopolitical and technological storm, there remains room for the whimsical and the human. Whether it is a ska cover of Rage Against the Machine, the discovery of just how far Stretch Armstrong can stretch, or the calming influence of customized ambient sounds, these moments of distraction are essential. Even a girl guiding a seal to perform tricks reminds us of the simple, non-algorithmic bonds that define the human experience.

<p><strong>Why is the race for rare earth elements vital for the future of AI?</strong><br>
Rare earth elements are necessary for the high-performance magnets and semiconductors that power AI hardware. Dependency on a single supplier, such as China, poses a significant risk to national security.</p>

<p><strong>What is the 'offensive AI' advantage mentioned by security experts?</strong><br>
Offensive AI refers to the use of machine learning to automate cyberattacks and hacking at speeds and scales that current human-led defensive teams cannot match.</p>

<p><strong>How is the Neanderthal DNA debate related to human evolution?</strong><br>
New theories suggest that what was thought to be interbreeding between Homo sapiens and Neanderthals might actually be an effect of population structure, potentially changing how we view our ancestral history.</p>

<p><strong>Are voice actors being replaced by AI in Hollywood?</strong><br>
Many voice actors are fighting the use of their voices to train AI models, as these models can then generate synthetic speech that mimics them, threatening their employment.</p>

Join the Conversation: Do you believe “humans in the loop” is a genuine safety measure or a psychological comfort? And as AI begins to mimic our creativity and voices, what is the one thing you believe will always remain uniquely human?

Share this article with your network and let us know your thoughts in the comments below.