New Mexico lawmakers are weighing the risks of increasingly capable artificial intelligence after an Anthropic researcher warned that AI companies were “gambling with our lives.”
Jacob Coxon, who has worked at both Anthropic and OpenAI, recently resigned from Anthropic over questions surrounding AI safety. In a warning that has drawn attention in Washington, Coxon said companies were “gambling with our lives.”
The warning comes after Anthropic disclosed Thursday that it had blocked attempts to misuse its AI in ways that could have supported the development of biological weapons.
The incident has renewed questions about how AI systems could be misused and whether additional safeguards are needed as the technology becomes more capable.
New Mexico Sen. Martin Heinrich, a Democrat, said the potential benefits of AI must be accompanied by attention to its risks.
“AI’s rapid advancement could deliver immense opportunities,” Heinrich said. “But this is a real warning, and we need to take it seriously.”
Heinrich called for “clear guardrails” to ensure AI is developed and used safely, ethically and in the public interest.
Sen. Ben Ray Luján, another New Mexico Democrat, also called for Congress to act.
“With AI becoming more powerful, it is vital Congress act swiftly to ensure guardrails and defenses are put in place,” Luján said.
Luján pointed to his bipartisan TEST AI Act, which would direct the federal government to build testbeds for evaluating AI systems. He said Sandia National Laboratories and Los Alamos National Laboratory support the work, which he described as having the technical expertise and computing resources needed for the effort.
Luján also said he has cosponsored legislation addressing transparency, risk assessment and accountability for the most powerful AI systems.
What is at stake?
At the heart of the debate is how to test and control AI systems as their capabilities grow.
The Anthropic incident provides a concrete example of the concern. According to the company, it blocked attempts to misuse its AI in ways that could have supported biological-weapons development.
That raises questions about how AI systems might be used by people seeking to carry out harmful activities—and about what safeguards should be in place to prevent such misuse.
The challenge lawmakers face is determining what protections are necessary while allowing AI research and development to continue.
Heinrich’s call for “clear guardrails” and Luján’s emphasis on testing, transparency, risk assessment and accountability point to the same central issue: how to make increasingly capable AI systems safer and more accountable.
The warning from Coxon and Anthropic’s disclosure has put that question into sharper focus for New Mexico’s congressional delegation.
The debate now facing lawmakers is not simply about what AI can do. It is about what safeguards should be in place when the technology is used—and how those safeguards can keep pace with its growing capabilities.
