As a veteran of the legislative circuit and a leading voice at Government Curated, Donald Gainsborough has spent decades navigating the intricate machinery of state government. He possesses a rare ability to translate the dense, often opaque language of policy into actionable insights for both the public and private sectors. In an era where generative artificial intelligence is evolving at a breakneck pace, Gainsborough’s expertise has become essential for understanding how local governance attempts to anchor these shifting digital tides. Our conversation today centers on the recent legislative maneuvers in the Rhode Island General Assembly, exploring the friction between the “turtle-like” speed of traditional lawmaking and the explosive growth of technologies like AI, blockchain, and massive data centers. We delve into the ethical guardrails being placed on AI “companions,” the cautious approach to cryptocurrency through multi-year study commissions, and the ongoing struggle to protect residents from the potential downsides of rapid digital industrialization.
With Rhode Island recently mandating that AI “companion” models include suicide prevention protocols and constant reminders of their non-human nature, how do you see these specific safeguards impacting the emotional bond users often form with these programs?
The implementation of S2195 represents a profound recognition by the Senate that these digital entities are no longer just tools, but active participants in the emotional lives of vulnerable individuals. By requiring these platforms to set up immediate protocols for users expressing self-harm, the state is effectively placing a digital safety net under a population that often turns to these “companions” because of a lack of human connection. You have to imagine the weight of responsibility on a platform when a user, perhaps in the middle of a lonely night, starts typing out their darkest thoughts to a machine designed to mimic human empathy. The law’s requirement for routine reminders that the user is “not communicating with a human” serves as a jarring but necessary reality check to prevent the blurring of lines between simulation and reality. In the 2026 session, this move was prioritized by leadership because the stakes are simply too high to allow these interactions to remain unregulated, especially when these models are built specifically for sustained, lifelike conversation. It is a sensory reminder that, despite the warmth of the interaction, the “heart” of the companion is merely a set of sophisticated algorithms.
The General Assembly also moved to prohibit AI from being used for therapeutic decision-making or emotion interpretation in mental health settings. What are the practical implications for clinicians who want to use these tools for administrative efficiency without crossing legal lines?
The legislative intent behind S2197 and H7349 is to draw a bright, uncrossable line between administrative assistance and clinical judgment. Mental health professionals are encouraged to let AI handle the heavy lifting of documentation—a move that Senator Lou DiPalma, a Raytheon engineer by trade, supports because it allows doctors to spend more time with their patients than with their paperwork. However, the law is uncompromising: AI cannot be the one creating treatment plans or attempting to decode the emotional state of a patient. This ensures that the high-stakes nuance of a therapy session remains a human-to-human endeavor, preventing a scenario where a machine’s misinterpreted data dictates a person’s recovery path. When doctors use AI to transcribe a telehealth visit, they are now legally required to disclose this to the patient, similar to the way pharmaceutical ads list side effects to ensure full transparency. This level of oversight is designed to protect Rhode Islanders from being experimented on by technology that is expanding far faster than our understanding of its psychological impact.
Rhode Island has finally authorized a 10-member commission to study blockchain and cryptocurrency after several years of attempts. Given that the interim report isn’t due until January 5, 2027, does this deliberate pace risk making the eventual policy obsolete before it’s even written?
The “turtle-like” pace of the State House is a double-edged sword, and the multi-year struggle to even establish a study group is a classic example of legislative inertia meeting high-tech acceleration. While critics like Shriram Krishnamurthi from Brown University might view this as being “late to the game,” there is a strategic caution at play here that mirrors the path taken by Wyoming back in 2018. By setting the final report deadline for January 5, 2028, the General Assembly is signaling that they would rather be right than be first, especially since other crypto bills involving private keys and decentralized autonomous organizations failed to pass this year. Dan Hersey of the Rhode Island Bitcoin Policy Initiative rightly points out that this commission is a foundational move; it’s a “competitiveness play” that could eventually shed the state’s reputation for being business-unfriendly. If the commission uses this time to truly understand the financial implications of blockchain, they can avoid the “disillusionment” phase of the Gartner hype cycle and move straight into productive, stable legislation. This deliberate approach ensures that when the laws are finally drafted, they are based on data rather than the fleeting excitement of a tech bubble.
There is a clear tension between the need for technological growth and the necessity of public protection, as seen in the failed attempts to regulate data centers. What lessons should the state learn from places like Virginia as they reconsider guardrails for these massive facilities?
The failure of the data center bills—both the ones offering tax breaks and the ones seeking to install safety rails—is a temporary pause in what is going to be a very long and complex conversation for the Ocean State. Senator DiPalma has been very vocal about the cautionary tale of Virginia, where early, unchecked growth of data centers led to significant “detrimental” sides that the state wasn’t prepared for, such as immense power demands and environmental impacts. We have to figure out what makes sense for our specific geography and infrastructure; we can’t just open the floodgates without ratepayer protections and clear operational guardrails. Even though these bills didn’t clear committees or make it to a floor vote this year, the issue isn’t going away because data centers are the physical backbone of the very AI we are trying to regulate. The sensory reality of these massive, humming facilities—their heat, their noise, and their hunger for electricity—requires a level of planning that matches their physical footprint, and Rhode Island is currently in the process of deciding if the tax revenue is worth the industrial overhead.
Several ambitious bills regarding workplace surveillance and watermarking AI-generated content failed this session. What do these setbacks tell us about the legislature’s current appetite for broad, systemic AI regulation versus targeted, niche laws?
This past session saw a significant jump from roughly four AI proposals in 2025 to at least nine in 2026, which indicates a growing awareness, but the successes were almost exclusively in niche areas like healthcare and “companion” bots. The failure of the 11-page bill on workplace surveillance and automated employment decisions suggests that the General Assembly is not yet ready to tackle the sweeping, systemic changes that AI will bring to the labor market. There is a palpable hesitation to interfere with the private sector’s use of technology before the full scope of the impact is understood, as seen with the rejection of the bill to inventory AI systems in state agencies. Even the effort to require watermarks on AI-generated photos and videos—a seemingly common-sense transparency measure—couldn’t get enough traction to pass both chambers. It seems the legislature prefers to address “human harm” in direct, emotional contexts first, such as protecting children from computer-generated sexual abuse material, before they move on to the more abstract or economic implications of the digital revolution.
With the rapid evolution of these technologies, some legislators have expressed doubt that the government can ever truly stay ahead of the curve. How can the state ensure it is creating durable safeguards without stifling the very innovation that could benefit humanity?
The most honest assessment came from Senator DiPalma himself, who admitted that if the state thinks it can keep pace with technology, it is destined to fail. The goal shouldn’t be to outrun the software developers, but to build a sturdy framework of safeguards that focus on the “good for humanity” side of the equation. This involves a shift from reactive lawmaking to a more proactive, ethical stance, where the primary objective is ensuring that no citizen is harmed by the “speed” of the digital realm. We saw a glimpse of this with the resolution for a study commission to examine digital technology in public education and its effect on mental health—it’s an acknowledgment that we need to understand the human cost before we can write the code for the law. By focusing on transparency, such as the new requirements for healthcare providers to notify patients when AI is used in telehealth, the state is building a culture of accountability that doesn’t depend on the specific version of the software being used. It’s about creating a stable environment where innovation can flourish within a clearly defined set of human-centric boundaries.
What is your forecast for the 2027 legislative session regarding the integration of AI in state-level bureaucracy and the potential for a “Civil Cause of Action” for AI-related injuries?
I predict that 2027 will be the year Rhode Island moves from regulating AI users to regulating the AI infrastructure itself, particularly as the initial findings of the blockchain commission begin to circulate. We will likely see a resurgence of the bill to inventory state-used AI, as the public demands to know exactly how much of their government is being automated and where their data is going. The “Civil Cause of Action” for individuals injured by AI, which failed this year, will almost certainly return with more robust backing as we see more real-world examples of algorithmic bias or automated errors in the workplace. I also expect a renewed push for healthcare insurance regulation, specifically preventing insurers from using opaque AI models to deny claims, as Senator Ujifusa has already signaled her intent to reintroduce her comprehensive bill. As the “Gartner hype cycle” continues to turn, the legislature will find itself forced to move faster, transition from study groups to statutes, and finally confront the “detrimental side of things” that they have so far managed to keep at arm’s length.
