The Future of AI Regulation: Debunking the Myths
- marcuspark
- Jul 8, 2025
- 5 min read
Updated: Sep 24, 2025
AI Is Already Subject to Business Regulations
The truth is, AI is already subject to an expansive web of legal rules, safety regulations, and doctrines. These laws may not have been written with AI in mind, but they apply all the same. As AI systems grow more sophisticated and integrated into decision-making processes in finance, healthcare, law, transportation, and beyond, the legal obligations and liabilities of those who build and deploy them become even more pronounced.
This article argues that the premise of a regulatory void is both incorrect and dangerous. More importantly, we assert that AI practitioners should not wait for new AI-specific regulations to start protecting themselves and their customers. Legal risk is already here, and avoiding it requires a fundamental understanding of tort law, fiduciary responsibility, negligence, and strict liability standards.
And as we will see, there's already a powerful solution: integrating human oversight directly into AI systems. Platforms like Pearl's Expert-as-a-Service MCP Server offer a practical way to comply with existing legal duties while preparing for the regulatory frameworks to come.
Congress Leaves Regulation on the Table
The proposed 10-year moratorium on AI regulation was based on two key assumptions: First, that the United States needed time to fully understand the implications of advanced AI technologies before enacting laws. And second, that AI companies would be better off if they were shielded from a patchwork of potentially conflicting state laws. Proponents argued that this approach would:
Encourage innovation without premature constraint
Offer regulatory clarity at the federal level
Help U.S. companies compete internationally
But Congress ultimately rejected this approach. By removing the moratorium, lawmakers affirmed that AI safety and accountability cannot be deferred. Delaying regulation doesn't make AI safer or more trustworthy; it simply delays accountability.
In practice, no such regulatory vacuum exists. Companies must still follow federal regulations, enforce rules, and ensure the safe and ethical use of their systems.
Tort Law Already Enforces Rules on AI
Take this example: A hedge fund manager uses an AI system to generate a new stock trading strategy. The model is trained on historical data, fine-tuned with proprietary signals, and deployed with little oversight. Within days, the strategy results in a $1 million loss. Is there any legal recourse?
Yes. And it's likely grounded in centuries-old principles of tort law.
According to Ayres and Balkin in their landmark paper The Law of AI is the Law of Risky Agents Without Intentions, AI systems should be treated like human agents operating without intent. This means the law holds developers and deployers of AI accountable under objective standards of care.
In such cases, three legal doctrines come into play:
Negligence: Did the developer exercise reasonable care in designing, testing, and deploying the AI?
Strict Liability: In high-risk domains, was the AI held to the highest standard of care, regardless of intent?
Duty of Care: Was there any reduction in responsibility due to the use of AI instead of a human actor?
As Ayres and Balkin argue, substituting an AI for a human does not reduce the duty of care. If an AI makes a contract, misstates information, or produces false financial analysis, the company using that AI is still fully responsible for the consequences.
Current Laws Already Assign Responsibility
One of the most compelling aspects of existing legal frameworks is their flexibility. Even without AI-specific laws, courts can assign liability in several key ways:
Respondeat Superior: Principals are liable for the actions of their agents. If your AI acts on your behalf, you are responsible.
Defective Design: Like any product, an AI tool can be found legally defective if it fails to perform safely under foreseeable conditions.
Failure to Supervise: Deployers can be liable for not adequately supervising or testing their AI systems before launch.
This means AI developers and their clients need to think not just about what their systems can do, but also what could go wrong—and who gets hurt when it does. In highly regulated industries like nuclear power plants or financial services, the stakes are even higher.
The Danger of 'Intention-Free' Agents
Many areas of law rely on the idea of intention (or "mens rea") to determine liability. But AI doesn’t have intentions. It doesn’t know, choose, or feel anything. So how can it be liable?
That’s precisely the issue. Legal scholars now propose that AI should be governed not by subjective intent, but by objective standards. This idea is gaining traction across regulatory and academic circles, including through the work of Margot Kaminski, who advocates for applying traditional agency and fiduciary concepts to AI agents.
As explained in The Human Factor in AI Regulation on MarkTechPost:
"Substituting an AI agent for a human agent should not result in a reduced duty of care."
What does this mean for practitioners? It means you must proactively ensure that your systems:
Are trained with high-quality, representative data
Include safeguards for edge cases
Include escalation mechanisms for human oversight
Real-World Accountability Isn't Theoretical
Let’s revisit our hedge fund example. If the AI-generated stock strategy was based on outdated or misleading data, and the developers failed to update or monitor it, courts may find them liable under negligence.
If the system was sold as a financial advisory tool but lacked any human expert review, it might violate fiduciary duty principles. If marketed recklessly or deployed in high-risk contexts, it could expose the company to strict liability.
This is not hypothetical. These cases are already surfacing in courts worldwide. AI-generated content has led to defamation, wrongful arrests, and financial fraud. In each instance, existing laws, not future regulations, have been used to pursue justice. In fact, the vast majority of AI deployments today are governed by overlapping safety regulations, business regulations, and rules enforced by governmental agencies.
How Pearl AI + Experts Offer Legal Resilience
Rather than waiting for regulation to catch up, AI builders should take the initiative to build human accountability into their systems. Pearl AI's platform offers exactly that.
Using the Model Context Protocol (MCP), Pearl enables AI agents to seamlessly invoke verified human Experts in real-time. Instead of replacing human judgment, Pearl enhances it. Here’s how it works:
MCP Integration: AI agents can discover and escalate to human Experts when confidence is low or the stakes are high.
Verification: All Experts are credentialed and peer-reviewed.
Real-Time Access: Legal, medical, financial, and technical Experts are available in minutes, 24/7.
This hybrid model is not just a best practice; it may be the safest way to deploy AI today.
Making Risk Reduction Your Default Design
If you want to stay ahead of both regulation and litigation, you need to treat AI risk the way airlines treat flight risk: Plan for failure. Build in redundancies. Assume things will go wrong and prepare for escalation.
That means:
Integrating human oversight
Logging decision pathways
Designing for auditability
Adopting ethical and inclusive training datasets
Committing to transparency and accountability at every step
Pearl’s MCP Server helps AI builders meet all of these needs without slowing down development. It makes trust scalable.
Conclusion: Don’t Wait for AI Law to Get Serious
The future of AI regulation is still wide open. Lawmakers rejected the idea of a decade-long pause. The regulatory conversation is very much alive.
The myth of a regulatory vacuum has been officially debunked. With regulation still on the table, companies must focus not on delaying the inevitable, but on designing AI systems with accountability from day one. Pearl’s AI + Expert platform gives you a head start. By embedding human insight into the heart of your AI systems, you not only increase quality and trust; you future-proof your business.
Comments