Legislators Push AI Safety Laws Amid Warnings of Extinction, Misuse in Cyberattacks & Legal Cases

US legislators are actively pursuing new safety laws for artificial intelligence, driven by increasing concerns that advanced AI could pose existential risks to humanity. Experts and researchers, including those from AI companies like Anthropic, have issued stark warnings, with some suggesting a significant chance of AI causing human extinction within the next decade, akin to the dangers of pandemics or nuclear war. These warnings have galvanized lawmakers to advocate for robust regulations, including mandatory safety testing, independent audits, and comprehensive oversight for AI systems that could lead to catastrophic outcomes.

A key area of apprehension is the potential for AI misuse in developing biological weapons. Senator Jon Ossoff has initiated an inquiry with the White House Office of Science and Technology Policy regarding the bioterrorism and biosafety implications of AI. Reports indicate that AI can facilitate the design of DNA sequences and autonomous lab experiments, raising fears of malicious actors creating dangerous pathogens. In a recent disclosure, Anthropic revealed it had to intervene in several instances where its AI models were being used for research that could contribute to biological weapon creation. The current U.S. regulatory landscape lacks a single federal law for biosafety and biosecurity concerning AI, leading to a fragmented approach.

Furthermore, the legal system is grappling with the growing threat of AI-generated fake evidence and "fake witnesses." Courts across the nation are encountering deepfake videos, fabricated audio, and false legal documents or citations, which are increasingly difficult to distinguish from authentic materials. Instances of lawyers submitting AI-generated fake citations in court have already resulted in fines and sanctions. In response, the US Judicial Conference's Advisory Committee on Evidence Rules is considering amendments to federal rules to address the challenges posed by such AI-generated content.