Publications: Daniel Castro

In China Daily, Daniel Castro argues that the U.S. and China face AI risks—like models enabling biological threats or cyberattacks—that are too great for either to manage alone, and can be mitigated through coordinated safety measures such as joint research, incident reporting, and red-team testing.

Fifty conflicting state AI laws create a fragmented, innovation-crushing patchwork, which federal preemption can solve by establishing a single, coherent national framework for AI regulation.

France’s criminal investigation into TikTok for imperfect content moderation sets a dangerous precedent that would chill lawful speech and push platforms toward overly restrictive policies.

The Federal Railroad Administration should approve the expansion of Automated Track Inspection. This proven technology enhances rail safety and efficiency, but political pressure, outdated regulations, and an inconsistent waiver process have stalled its deployment.

Senator Blackburn’s call to shut down Google’s AI over a false claim is misguided, as the error came from a small, open developer model not designed for factual accuracy, and the incident does not demonstrate political bias or systemic failure.

The UK’s refusal to formally designate China as a national security threat has undermined its ability to prosecute espionage, leaving its technology and innovation sectors vulnerable to Chinese infiltration and economic coercion.

The UK Online Safety Act aims to protect children online but its vague rules and strict enforcement have led to over-censorship, threatening legitimate communities, and Parliament should clarify content definitions, allow remediation periods, and require judicial review to fix these issues.

In the short term, Congress should move quickly to reauthorize CISA 2015 and fund the agency’s operations, even if the overall government funding fight continues. Then, to keep CISA fully operational in the future, Congress also should create a dedicated funding stream so the agency doesn’t have to rely so heavily on appropriations.

California’s proposed SB 7 would heavily restrict employers’ use of AI in workplace decisions through onerous notice, transparency, and appeal requirements, creating redundant regulations that discourage beneficial AI adoption and ultimately harm both workers and businesses.

Lawmakers in Montana, New Hampshire, and Idaho are advancing “right to compute” laws to protect individuals and businesses from limits on their ability to use computational tools and AI systems.