Detailed Analysis
The Neuron Daily's headline "You're either Jeremy or you're cut" encapsulates a defining tension in the 2026 AI labor market: the bifurcation between workers whose expertise in artificial intelligence renders them indispensable and those being systematically eliminated to fund massive AI infrastructure buildouts. The phrase references Jeremy Kahn, Fortune's AI editor and co-chair of the Brainstorm AI conference, as an archetype of the "safe" professional — someone whose deep, specialized understanding of AI positions them above the automation threshold. The immediate catalyst for this framing is Meta's announcement of approximately 8,000 job cuts tied directly to its $72 billion AI investment strategy, a move that exemplifies how major tech corporations are reallocating human capital budgets toward compute, model development, and infrastructure rather than headcount.
The broader labor dynamics the article invokes are supported by Anthropic's own data, which suggests AI systems currently expose 94% of tasks to automation but see only 33% actual usage — a gap that has so far produced wage compression and suppressed hiring of younger workers (down 14%) rather than outright mass unemployment. This nuance complicates the popular binary of "jobs lost vs. jobs saved," suggesting the more immediate economic injury is structural: stagnant or declining wages and narrowing entry points for early-career workers. The "Jeremy or cut" framing maps cleanly onto this reality, as domain expertise in AI — editorial, analytical, or technical — has become a form of labor arbitrage protection that most workers do not possess.
Anthropic's recent developments deepen the article's context considerably. The April 2026 announcement of Claude Mythos Preview, internally codenamed Capybara, represents a significant capability escalation: the model is reportedly able to identify thousands of high-severity software vulnerabilities at scale. The decision not to release Mythos publicly, instead channeling access through Project Glasswing — a consortium of major technology firms coordinating responsible disclosure and patching — reflects Anthropic's continued emphasis on staged, safety-conscious deployment. Government notifications and pre-briefings for security experts further signal that Anthropic and regulators regard this generation of models as qualitatively different from predecessors, warranting institutional coordination rather than open release.
Simultaneously, emerging research on AI "mirages" — reasoning traces that may not accurately reflect the actual internal computations driving a model's outputs — adds a layer of epistemic uncertainty to the accelerating deployment of systems like Claude. If explanations generated by models diverge from their underlying "neuron" activations, the interpretability frameworks that safety researchers and regulators rely on are potentially unreliable, raising foundational questions about oversight. This problem intersects with the recent leak of code surrounding the agentic harness for Claude Code, which exposed details of how Anthropic structures autonomous task execution — a disclosure that underscores the security risks accompanying increasingly capable and widely deployed AI agents.
Taken together, the article reflects a moment of profound industrial restructuring in which capital concentration, capability acceleration, and workforce displacement are advancing in tandem. Google's commitment of up to $40 billion in cash and compute to Anthropic cements the company's position as a central infrastructure node in the AI economy, even as questions about model transparency, security, and labor displacement remain unresolved. The "Jeremy or cut" formulation is less a career tip than a structural diagnosis: in an economy reorganizing around AI capabilities, the distance between indispensability and expendability is narrowing rapidly, and the credentials that once conferred job security are being rapidly repriced or rendered obsolete.
Read original article →