Daniel Schwarcz

This framework draws on principles of consumer financial regulation and lessons from the EU’s new AI Act. It seeks to balance the promise of AI-driven financial advice with the need to protect consumers from significant harm. Check it out!

Our solution is a dual regulatory approach: (1) a licensing requirement for robo-advisors that use generative AI to match consumers with products, and (2) heightened duties of care and loyalty for all robo-advisors.

We argue that the current U.S. regulatory framework is not up to the task. Existing rules fail to prevent AI-enabled robo-advisors from providing conflicted, inaccurate, or manipulative advice on a large scale.

Generative AI can make customized financial advice widely available. But because it mimics human advisors so convincingly, it also creates serious risks that consumers will be nudged into costly or inappropriate products

I’m excited to announce the publication of Regulating Robo-Advisors in an Age of Generative Artificial Intelligence in Volume 82 of the Washington & Lee Law Review . The article is co-authored with @notthattombaker.bsky.social & Kyle Logue. lawreview.wlulaw.wlu.edu/regulating-r...

SSRN; "Obamacare for Homeowners Insurance: Fixing America's Broken Insurance Markets in a Time of Climate Change," 49 Harvard Environmental Law Review (2025); Prof. Dan Schwarcz

@danielschwarcz.bsky.social’s article, "Obamacare for Homeowners Insurance: Fixing America's Broken Insurance Markets in a Time of Climate Change," 49 Harvard Environmental Law Review (2025), is now available on @ssrn.bsky.social.
z.umn.edu/aq4n.

Check out the broader law review article, The Limits of Regulating AI Safety Through Liability and Insurance: Lessons From Cybersecurity. on which our Lawfare piece is based here: papers.ssrn.com/sol3/papers....

🚨 New on @lawfaremedia.org -- Josephine Wolff & I argue that the push to impose liability on AI companies won’t make AI safer—because insurers, not courts, will shape how liability plays out. And history from cyber insurance shows insurers don’t incentivize real safety.

8/ Without bold reform, insurance instability could cascade into a wider housing and financial crisis. With reform, it can be harnessed to build resilience. Full argument in Harvard Environmental Law Review: journals.law.harvard.edu/elr/wp-conte...

7/ Done right, this approach wouldn’t just stabilize insurance markets. It would make coverage a tool for climate adaptation: rewarding resilient building, discouraging risky development, and protecting vulnerable communities

6/ I argue we need an “Obamacare for Homeowners Insurance.” That means: (i) Minimum coverage standards (including climate risks like floods), (ii) anti-discrimination rules so insurers price fairly, (iii) State-run exchanges for managed competition, & (iv) Progressive subsidies.

5/ The crisis mirrors health insurance before the Affordable Care Act. Back then, broken state markets left millions uninsured. The ACA offered a template: federal standards, state implementation, broad coverage rules, and progressive subsidies.

4/ State fixes have failed. Florida blames lawsuits; California doubles down on rate regulation. Both approaches misdiagnose the problem—and both have fueled more instability, not less.

3/ Climate change is the main driver. Wildfires, floods, hurricanes, and severe storms are making coverage unaffordable and unavailable. Even states once thought “safe” are now facing double-digit premium hikes and insurer exits.

2/ U.S. homeowners insurance is breaking down. Insurers are pulling out, premiums are spiking, and millions are left without reliable coverage. This crisis isn’t just about insurance—it threatens housing markets, financial stability, and climate resilience.

1/ Just published in the Harvard Environmental Law Review: “Obamacare for Homeowners Insurance: Fixing America’s Broken Insurance Markets in a Time of Climate Change.” journals.law.harvard.edu/elr/wp-conte...

Bottom line: Liability and insurance can’t carry the regulatory load. Effective AI safety governance requires proactive regulation—setting ex ante rules for design, testing, and deployment of AI systems. Check out the full argument here: papers.ssrn.com/sol3/papers....

If liability insurers cannot advance AI safety through pricing or other strategies, then liability becomes a limited regulatory tool, since most firms are already insured against AI-related risks through general liability coverage and are likely to remain so in the future.

This doesn’t mean insurers have no role to play in mitigating narrow AI risks (e.g., performance guarantees) or in helping firms weather broader AI liability. But expecting them to regulate AI safety is a mistake.

Without strong ex ante rules, liability insurance won’t promote AI safety. At best, it will pool losses after accidents. At worst, it will distort incentives, encourage secrecy, and lull policymakers into false confidence.

AI safety presents even greater challenges for insurers than cybersecurity, due to the scarcity of reliable loss data, the rapid evolution of AI systems, the technical complexity of ensuring safety, and the fact that AI-related risks are embedded throughout business operations.

History tells us otherwise. Cyber insurers were once hailed as “private regulators” who would make firms safer from cyber risk. Two decades later, they have struggled to price risk in ways that meaningfully encourage cybersecurity or otherwise advance it.

Many argue that if AI firms are liable for harms, insurers will price that risk & push them toward safer practices. Sounds elegant. But there’s a problem… it assumes insurers can measure those risks well enough to incentivize care, or otherwise affirmatively promote AI safety.

🚨 New paper alert! Josephine Wolff & I argue that efforts to regulate AI safety through liability and insurance are likely to fail—just as they largely did with cybersecurity. Download The Limits of Regulating AI Safety Through Liability & Insurance: papers.ssrn.com/sol3/papers....

Had a great time presenting at the Law-Following AI Workshop at Cambridge. Drawing on the track record of cyber insurers, the project argues that liability insurance is likely to limit the effectiveness of liability as an AI regulatory tool. Will share the paper publicly soon!

Fair enough, but homeowners isn't much of a long-tail line. Even if earned-to-incurred were at more like 50 or 60%, that would not be indicative of a competitive market in homeowners, one would think, given limited non-claim expenses, right?

Can someone please explain to me how loss ratios in the NH homeowners insurance market have consistently hovered between 30-40% in recent years? Doesn't that seem like clear evidence that homeowners insurance markets are broken? www.insurance.nh.gov/sites/g/file...

The Eighth Edition of Insurance Law and Regulation (Abraham & Schwarcz) is now available. Updates address key recent issues in insurance, such as the COVID-19 pandemic, opioid crisis, and gen AI. faculty.westacademic.com/Book/Detail?...

This finding aligns with my own work, papers.ssrn.com/sol3/papers...., which warns that lawyers who rely on AI to complete cognitively demanding legal tasks may struggle with related work—such as engaging effectively with clients or judges, or integrating insights across discrete tasks.

Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task This study explores the neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed thr...

A new essay, arxiv.org/abs/2506.08872, presents evidence that using AI can impair users’ cognitive understanding of the underlying material.