A Regulatory Vacuum Fills Itself

The US government has yet to enact broad federal legislation governing AI safety, leaving a gap in how advanced systems are tested and certified. Into this space, independent research organizations specializing in AI model evaluation have moved quickly, offering external assessments that governments and companies alike have increasingly turned to.

Among these groups, METR (Model Evaluation & Threat Research) has become a visible name in the field, known for its work evaluating frontier models. The rise of such third-party evaluators suggests that safety assessments are being shaped largely outside of formal legislative processes.

Implications for Korean Industry

For Korean companies developing large language models (LLMs) — including players such as Naver, Samsung, and Kakao — the trend points toward growing reliance on independent evaluation bodies, particularly when entering the US market. The absence of a US federal regulator means the standards these private organizations set may effectively function as benchmarks.

No binding US federal AI framework has been confirmed to date, and the extent to which third-party evaluations will be formalized into government policy remains uncertain.

Conclusion

The claim that third-party evaluation organizations are rapidly rising amid a US government AI regulatory gap is consistent with the available evidence. Verdict: Mostly True