AI And Cybersecurity: The New Administration's 100-Day Reckoning

At the RSAC 2025 conference this week, Snyk hosted a timely panel titled “The First 100 Days: How AI, Policy & Cybersecurity Collide,” featuring an all-star lineup: Jen Easterly, former CISA Director; Nicole Perlroth, former journalist and partner with Ballistic Ventures; Sumit Dhawan, CEO of Proofpoint; and Peter McKay, CEO of Snyk. Moderated by Axios cybersecurity reporter Sam Sabin, the conversation examined the early signs of disruption and dysfunction—and what it all means for software security, national defense, and innovation.

The discussion was grounded in new findings from a Snyk-commissioned CISO survey, which revealed stark concerns about AI-generated threats, fragmented regulation, and eroding trust between the public and private sectors.

A Surge of AI-Driven Attacks—and a Growing Crisis of Trust

Since January, 70% of surveyed CISOs reported experiencing a cyberattack involving AI. Panelists noted that organizations are rapidly embracing AI to increase productivity, but often without properly considering security implications.

This rush to adopt AI is creating a widening gap between innovation and risk management. At the same time, nearly all CISOs surveyed expressed concern that AI-generated code may be introducing hidden vulnerabilities, suggesting a dangerous disconnect between perceived readiness and the evolving threat landscape.

Peter McKay observed, “Everybody is just focused on productivity... just get the benefits of AI and we’ll figure out security later,” highlighting the widespread rush to adopt AI tools without sufficient safeguards.

Regulatory Uncertainty and Talent Flight Undermine National Security

The panel addressed the impact of federal workforce reductions and policy reversals, including the rollback of Biden-era AI executive orders. Former CISA Director Jen Easterly described the loss of technical talent from government agencies as damaging to national cyber readiness.

The panelists noted that reported loyalty requirements for federal cybersecurity personnel could further erode morale and independence. Concerns also extended to international partnerships, with reports that allied nations are beginning to limit intelligence sharing with the U.S., reflecting declining trust in the current administration.

Developers Are Racing Ahead of the Guardrails

AI is drastically accelerating software development cycles, but this rapid pace is straining traditional security frameworks. Panelists highlighted how internal pressure to innovate often overrides caution, leading to insufficiently vetted tools and code.

They stressed the importance of integrating security from the outset rather than as an afterthought, and called for secure-by-design practices to become standard. Without these safeguards, AI tools that can prevent threats may also be exploited to cause harm.

Public-Private Collaboration at Risk

Speakers emphasized that recent setbacks in federal cybersecurity leadership and policy risk reversing years of progress in public-private cooperation. The collaboration that once enabled a strong collective response to cyber threats—most notably in Ukraine—is now showing signs of strain.

Reflecting on current limitations, Dhawan admitted, “We haven't cracked the code, and we're really hoping for that form of private, public sort of collaboration.”

Several panelists expressed concern that trust is weakening on both sides, with private companies unsure about their role in threat reporting and mitigation, and government agencies losing key channels for visibility. CISA’s partnership-enabling authorities, such as CPAC, were cited as critical tools that are currently on hold.

Policy Wish Lists from the Experts

When asked what single change they would make if given a “magic wand,” panelists offered a range of pragmatic solutions. Proposals included mandating secure-by-design standards for consumer-grade routers—long a weak link in infrastructure security—and launching a national effort to clean up the open source codebase that underpins most modern applications.

Others called for harmonized, standardized AI development regulations to prevent a patchwork of conflicting state laws. There was also strong support for a software liability regime tied to demonstrable secure development practices, as well as the use of AI to refactor legacy code written in memory-unsafe languages.

As his top policy wish, McKay advocated for a national effort to improve software security at the source: “If we all just focused on how we can just clean up open source code, we would have been in a better place.”

An Urgent Need for Coordination

A unifying theme throughout the discussion was the urgent need for coordination—across sectors, agencies, and borders. The convergence of rapid AI adoption, regulatory rollbacks, and mounting cyber threats is creating a perfect storm. Industry leaders stressed that security cannot be an afterthought, and that public trust and international cooperation hinge on transparency, integrity, and mutual accountability. The panel concluded with a call to preserve the principles of trust and collaboration that once underpinned America’s cyber defense strategy—and to ensure those values guide policy moving forward.

Easterly closed with a reflection on her time at CISA and how that should serve as a guiding light moving forward: “We built trust and catalyzed trust and collaboration, and we did it with integrity, we did it with humility, we did it with transparency, and we did it with character. And that's what you all should demand from your government.”

The above is the detailed content of AI And Cybersecurity: The New Administration's 100-Day Reckoning. For more information, please follow other related articles on the PHP Chinese website!