Critics and safety advocates warn that the current pace of AI development is creating a dangerous environment where the risks to national security and public safety are not being adequately addressed. The shift toward agentic AI—systems that can perform tasks and make decisions autonomously—introduces a level of complexity that existing security frameworks are ill-equipped to handle. Skeptics argue that when companies prioritize speed and competitive advantage over safety, they create systemic vulnerabilities that could lead to catastrophic failures, including large-scale cyberattacks, the amplification of fraud, and the compromise of critical infrastructure.
There is a growing concern that the industry is underinvesting in safety research while over-relying on the assumption that future AI will be able to solve the problems created by current models. This 'move fast and break things' mentality is particularly alarming when applied to technologies that could be exploited by hostile actors to perpetrate deep repression or gain military superiority. Accountability-focused observers point out that without mandatory standards and rigorous, independent third-party evaluations, there is little incentive for companies to slow down or invest in the necessary safeguards. The lack of visibility into how these models are trained and deployed remains a significant blind spot for regulators and the public alike.
For these critics, the solution lies in stronger oversight and a fundamental shift in how AI risks are managed. They advocate for mandatory security standards, greater transparency in model development, and a more cautious approach to the deployment of powerful, general-purpose AI systems. By treating AI safety as a core requirement rather than an afterthought, the industry can prevent the most severe outcomes. The stakes are simply too high to leave the security of our digital and physical systems to the voluntary actions of companies driven primarily by market competition.