AI Competition Accelerates Across Military, Jobs, Web and Security

American college graduates are entering a job market gutted by the very technology that Silicon Valley billionaires promise will transform our lives for the better. The underemployment rate for recent grads has hit 42.5 percent, the worst since the pandemic, as artificial intelligence systems take over tasks that once provided entry-level experience. Young people describe a soul-crushing grind of sending out hundreds of applications only to be ignored or instantly rejected by algorithms that increasingly decide who gets a foot in the door.

Gillian Frost, a 22-year-old at Smith College set to graduate in May with a degree in quantitative economics, has applied to more than 90 jobs since September. She spends weekends buried in applications, has been ghosted by a quarter of them, and received automated rejections from over half. A handful of interviews led nowhere, with employers failing to even send basic rejection notices. “I feel helpless,” she told reporters. “No one seems to know how best to prepare due to the unique conflux of events occurring.” Frost is not alone. Jeff Kubat, 31, returned to school for a master’s in accounting after eight years in the workforce and now finds even those specialized roles evaporating.

This is not some natural evolution. It is the direct result of AI systems aggressively inserted into workplaces, often by the same companies that lecture us about ethics and progress. While graduates scrape by, the AI firms themselves are strip-mining the foundation of the internet itself. New data from Cloudflare, which powers about 20 percent of the web, reveals a startling imbalance. These companies send out automated bots to crawl and ingest content at massive scale but send almost no traffic back to the sites that created it.

Anthropic stands out as the most voracious. The company, led by Dario Amodei and frequently praised for its supposed commitment to responsible AI, has a crawl-to-refer ratio of 8,800 to 1. For every single time it refers a user back to a website, its bots crawl that site 8,800 times. OpenAI follows with a still lopsided 993 to 1. By comparison, more traditional players like Microsoft, Google, and DuckDuckGo operate far closer to balance. The grand bargain that built the web, creators produce content and technology companies help users find it, is being shredded. AI chatbots keep users inside their platforms, starving the original publishers of traffic, advertising revenue, and relevance. The little guys who built the internet are being harvested to train machines that will eventually replace them.

Even large enterprises are growing wary. A noticeable gap has opened between the frontier AI models controlled by a handful of powerful companies and what businesses actually feel comfortable using. OpenAI and Anthropic insist they do not train on enterprise customer data, yet both have faced repeated legal challenges over copyright violations and other scandals. Companies with sensitive intellectual property or customer information are increasingly reluctant to hand it over to these black-box systems. The result has been a surge in open-weight models from Google, Microsoft, Alibaba, and Nvidia that enterprises view as more controllable and trustworthy for real work. What was once dismissed as hobbyist technology now looks like serious competition for the closed systems pushed by the biggest names in AI.

This technological upheaval is not confined to the civilian economy. The same capabilities are fueling an international arms race that carries echoes of the early nuclear age. In September, China put on a military parade featuring autonomous drones capable of flying alongside fighter jets. President Xi Jinping hosted Vladimir Putin and Kim Jong-un to watch. American officials were alarmed enough to accelerate domestic programs. Anduril Industries recently began production of its AI-backed Fury drone three months ahead of schedule at a new Ohio factory. Russia and China appear to have moved faster on certain unmanned systems, forcing the United States to play catch-up in technologies that could one day select and engage targets with minimal human oversight.

The pattern is clear. A small group of powerful interests in Silicon Valley and Washington are charging ahead with artificial intelligence under the banner of national competitiveness and futuristic benevolence. The costs fall on everyone else: college graduates locked out of careers, independent websites stripped of value, businesses forced to choose between privacy risks and falling behind, and a world inching closer to automated weapons systems whose full implications few seem prepared to confront. The AI revolution is happening at accelerating speed. Whether it serves the broad interests of the American people or simply concentrates power and profit in fewer hands remains very much an open and troubling question.