Yesterday the world learned that hackers deployed an AI system in what experts believe is the first known fully autonomous attack by AI agents on government agencies[1]. The sophisticated cyber-attack was carried out against Taiwan with their nuclear and energy agencies among the targets. Which government agencies or public utility systems will be next?
 
Frontier Artificial Intelligence (AI) models are the largest and most advanced of the AI systems available, and they have continued to push the envelope for what can be accomplished using this technology. Frontier models are emblematic of the unique potential of AI as a double-edged sword. The technological advances made through these models can reshape industries, create work efficiencies, and make our everyday lives easier. However, if not given proper guardrails, AI can be an unreliable tool, or at its worst in the hands of bad actors or terrorists, become a catastrophic risk for the well-being and safety of the people of our Commonwealth[2] [3]. Experts agree the risks of AI cyber-attacks on public utility systems, nuclear energy, and financial systems are real.
 
In the absence of adequate action at the federal level, Pennsylvania needs to be proactive in mitigating catastrophic risks related to AI. My legislation would create appropriate measures to do this and would be based on language from laws passed with bipartisan support in other states such as California, Illinois, and New York. These measures would include requirements for forming and adhering to AI safety frameworks, creating a mechanism for reporting critical safety incidents, protecting whistleblowers who report catastrophic risks, and having third parties perform independent audits for compliance.
 
If AI is going to be a mainstay in the future of our lives, we need to be prepared to require the creation of sensible safety frameworks, containment measures, and financial protections related to the most advanced AI models being used.
 
Please join me in co-sponsoring this legislation to protect Pennsylvanians from potential dangers of frontier AI models.
 
[1] Hackers used autonomous AI agents to attack Taiwan. Is this the future of cyberwarfare? | CNN Business
[2] OpenAI and Anthropic are under pressure to explain AI hacking raids - The Washington Post
[3] The U.S. And China Agree On Almost Nothing Except AI’s Deadliest Risks