Voluntary safety pledges and calls to pace frontier AI development only bind compliant organizations, while malicious actors an adversary nations freely exploit open-source models. AI-enabled cyberattacks leverage familiar vulnerabilities—such as unpatched flaws and reused credentials—at machine speeds that vastly outpace human response times. The fundamental asymmetry in cybersecurity remains heavily tilted toward attackers, who only need one entry point while defenders must be correct every time. Despite surging cybersecurity budgets and high incident rates, a majority of organizations have yet to make AI central to their defense operations. Experts argue that rather than waiting for AI development to slow down, defenders must enforce strict security fundamentals and match adversary speeds using AI tools.

Full commentary : Voluntary restraint from major labs does little to halt bad actors, forcing cyber defenders to match the relentless speed of machine-driven attacks through rigorous fundamentals.