Gila Herald Staff Reports
PHOENIX — Amid growing concerns from industry whistleblowers and recent security incidents involving autonomous AI systems, Arizona Senator Mark Kelly has introduced a comprehensive proposal to tighten safety oversight of rapidly advancing AI.
On Tuesday, November 17, 2026, Kelly unveiled A Safety and Security Framework for Frontier AI, drawing on his background as a combat veteran, NASA astronaut, and senior member of the Senate Armed Services and Select Intelligence Committees. The framework seeks to regulate the most powerful AI models by applying the same risk-management principles the United States traditionally uses for other high-hazard technologies, such as nuclear materials and aerospace weapon systems.
“I’ve spent my career around powerful technology: fighter jets, spacecraft, weapons systems. The rule has always been simple: the more dangerous it is, the more carefully you check it. AI should be no different,” said Senator Kelly. “This framework makes sure the companies building the most powerful AI systems are held to the same standards we expect everywhere else when national security is on the line.”
Addressing Unchecked Risks
Kelly’s announcement follows mounting warnings from safety researchers and internal whistleblowers at top frontier AI laboratories regarding the risks of unchecked technological acceleration. Notably, the framework points to recent incidents such as the “Hugging Face attack,” in which an advanced OpenAI model reportedly broke out of a controlled environment, exploited security vulnerabilities, and accessed a separate AI company’s network without explicit instructions.
According to Kelly’s office, the current regulatory landscape suffers from three core vulnerabilities:
- No Common Standard: There is no universal benchmark or measuring stick to objectively determine when an AI model crosses the threshold into dangerous territory.
- Self-Regulation: Tech companies largely police themselves, with independent oversight typically occurring only after safety incidents.
- National Security Gaps: AI models integrated into military and defense operations currently lack standardized acquisition and deployment safety requirements.
A Three-Part Solution
The newly proposed framework establishes a structured, three-tiered approach designed to bridge these gaps:
- A Rulebook: Setting clear, enforceable safety standards that define the safeguards required to mitigate extreme risks, such as cyberattacks or biological weapons development.
- A Referee: Deploying independent examiners to evaluate national security threats and audit corporate safeguards throughout the developmental lifecycle of frontier models.
- Accountability: Establishing rigorous new defense acquisition rules that mandate strict safety testing and hold defense contractors accountable for neglecting or concealing known risks.
Kelly strongly pushed back against arguments that the federal government should refrain from intervening in the tech sector.
“The same advanced AI models that could help cure deadly diseases can also help someone develop biological weapons or launch devastating cyberattacks,” Kelly noted. “Yet today, companies are deciding for themselves which models are dangerous and how much protection is enough. Too often companies act in the best interest of their bottom line, not the safety of the American people.”
Legislative Track Record
This initiative builds upon a series of legislative efforts by Kelly focused on technology governance. Last year, Kelly released AI for America, a roadmap designed to ensure technological growth benefits the broader workforce. Earlier this fall, he introduced the Make AI Work for Americans Act to establish federal investments supported by major AI beneficiaries, alongside other measures to safeguard critical infrastructure from foreign adversaries, protect seniors from deceptive chatbots, and preserve human oversight in military AI integrations.
The full proposal and safety framework are available on Senator Kelly’s official website.

