On Thursday, the House passed H.R. 1: the “One Big Beautiful Bill Act.”  Consistent with President Trump's executive order aimed at removing barriers to American leadership in artificial intelligence, the Bill seeks to clear the path for more AI development and deployment in the US. It would put a ten-year moratorium on the enforcement of any state or political subdivision thereof enforcing any law or regulation regulating AI models, AI systems, or automated decision systems. Effective removal of state-level AI legislation would let AI development go effectively unchecked, absent oversight at the federal level.  The Bill would also appropriate funds to the Department of Defense, Department of Commerce, and the U.S. Customs and Border Patrol for deployment of AI systems.  It would also require the implementation of AI tools for the purpose of reducing and recouping improper payments under Medicare.  There is a lot more to unpack in the Bill, but one simple takeaway is that it is pro-AI.  It removes state-level AI guardrails while increasing government spending on AI. 

It's no surprise that IP strategies need to evolve as AI becomes more powerful and its deployments reach deeper into industries and geographies.  I recently wrote about how rescission of the AI diffusion rule will impact IP, and trade secrets in particular.  There, I focused on the likely devaluation of trade secrets that could be reverse engineered or independently created with increasingly powerful AI capabilities and increased valuation of trade secrets that cannot be created with AI.  

The rapid advances in AI cyber capabilities also impact IP strategy.   Anthropic reported in March about rapid improvement in offensive cyber capabilities of AI models, both on their own and when equipped with software tools built by cybersecurity researchers.  The Big Beautiful Bill's appropriation of a quarter billion for the expansion of U.S. Cyber Command artificial intelligence lines of effort reinforces the gravity of AI cyber capabilities.  Those capabilities, both offensive and defensive, impact both the battlefield and the boardroom.  What constitutes "reasonable measures” to protect trade secrets will evolve with the times.  And even with increased security efforts, trade secret theft surely will increase with increased AI capabilities.   

So what should trade secret holders do?  They should reevaluate their trade secrets and protection measures to make sure they are taking appropriate steps.  They should consider whether they need to enhance their digital security to sufficiently protect against modern threats. They should also review their access controls and compartmentalization of data.  While minimizing access to trade secrets has long been important, limiting access to data can also reduce the attack surface potentially exposed to cyber capabilities.  Employee training should also be updated in view of AI-driven attacks, such as the potential for more convincing phishing attacks or deepfakes.  Employees should also be trained on the risks of putting secret information into AI systems and the unique security risks that come with implementing large language models, such as prompt injection attacks.

Traditional security measures may not be “reasonable” in an era of IA-powered threats. Trade secret holders should evaluate appropriate measures to protect their trade secrets to minimize risk of misappropriation in the first place and to be prepared to show the reasonableness of their protection measures should their trade secrets be misappropriated and enforcement become necessary.