Loyola Community Raised Recursive AI Concerns

Faculty and students voiced risks regarding models that rewrite their own code without human intervention.

Updated on Sept. 23, 2026 in Artificial Intelligence

Bold flat-color editorial illustration of a microprocessor chip, representing autonomous code evolution. — Loyola University students and faculty have voiced concerns regarding the development of autonomous, recursive AI models capable of self-modifying code without human intervention. AI Illustration. Upload story photo >

Live Poll

Should the federal government implement stricter regulations on the development of artificial intelligence?

Loyola University faculty and students have expressed concerns over the development of recursive self-improvement artificial intelligence models. These systems theoretically possess the ability to rewrite their own code to evolve without human intervention.

Why it matters

The discourse highlights the tension between rapidly advancing AI capabilities and concerns regarding bias, career security, and human agency. As models evolve to self-correct, the current federal landscape has prioritized national uniformity over local regulation.

Recursive self-improvement AI refers to algorithms designed to rewrite their internal code to correct errors or enhance performance. These systems are typically trained on vast datasets aggregated from public websites and literature.

The players

Jacob Coxon

A former researcher at Anthropic, an AI safety and research company that develops large language models.

Donald Trump

The current President of the United States who issued an executive order regarding national AI regulation.

The details

Recursive models function by analyzing their own architecture to perform iterative code refactoring, theoretically increasing their intelligence over time. In contrast to static models that require human engineers to perform updates, these systems automate the development lifecycle to optimize their output. This transition toward autonomous code generation poses significant challenges to the static nature of existing training data, which often contains racial and social biases that could be amplified during self-modification.

2025: President Donald Trump signed an executive order blocking state-level AI regulations.
September 9, 2026: Former Anthropic researcher Jacob Coxon discussed recursive AI safety risks.
September 22, 2026: Opinions from Loyola University faculty and students were published.

The Tech Race

While the European Union has implemented a comprehensive AI Act to govern development, the United States has shifted toward federal preemption of local laws. This move highlights a growing divide in international governance regarding the risks of autonomous, self-improving systems.

The federal move to block state-level regulations ensures that AI policy remains centralized across the United States. Residents and workers should watch for how this shift affects emerging workforce protections and local tech oversight policies.

The takeaway

The gap between rapid technical advancement and current regulatory law remains wide, with the TAKE IT DOWN Act serving as the only federal measure. Observers should track potential new federal guidelines that may supersede current directives regarding self-modifying software.