Across the last quarter, the scholarship collected in AI Law Blawg has tended in two directions at once. One strand has become more concrete, institutional, and operational. Papers on algorithmic collusion, AI-driven cybersecurity, copyright training rules, data-sharing remedies, clinician oversight, and legal reasoning ask how existing legal regimes can be adapted to govern AI systems already embedded in markets, professional practice, and organizational decision-making. Their emphasis is less on broad principles than on administrable mechanisms: how to detect collusion, allocate access, structure liability, or assign oversight burdens.

A second strand continues to frame AI as a larger challenge to legal and institutional order. Work on legal alignment, superintelligence, institutional decay, and future personhood is less concerned with immediate implementation than with identifying pressure points at the level of legal architecture, legitimacy, and long-run governance.

What stands out is not merely the coexistence of both strands, but the widening distance between them. The more grounded work increasingly asks what existing institutions can absorb, while the more ambitious work asks whether existing institutions are conceptually adequate at all. If this pattern continues, a central question in the next phase of the literature will be whether these strands begin to converge, or whether the most useful work remains the scholarship that stays closest to concrete legal mechanisms and institutional constraints.

Some of the strongest recent work treats AI not as a set of isolated use cases, but as part of broader competitive and security systems. Sean Norick Long’s paper on autonomous algorithmic collusion, Nicholas Nugent’s on generative cybersecurity, Giovanna Massarotto’s on data monopoly remedies, and Michal Lavi and Hadar Jabotinsky’s on deepfakes in financial markets all share an important feature: they move beyond generic concern and toward structured legal responses to system-level problems. Taken together, they suggest a literature becoming more attentive to coordination, allocation, adversarial threat, and market design as recurring legal problems rather than exceptional ones.

• Sean Norick Long, The Mirror Test for AI Agents: A Path to Regulate Autonomous Algorithmic Collusion
• Nicholas Nugent, Generative Cybersecurity
• Giovanna Massarotto, Algorithmic Remedies for Google’s Data Monopoly
• Michal Lavi & Hadar Yoana Jabotinsky, Seeing is Believing? Deepfakes in Financial Markets (44 Cardozo Arts & Ent. L.J. 55 (2026))

The more abstract strand of recent scholarship remains active, but often at a different level of ambition and specificity. Work on legal alignment, superintelligence, institutional destruction, juridification, and future personhood continues to frame AI as a challenge to legal order, legitimacy, and institutional survival rather than as a problem of bounded legal design. This work raises important long-run questions, but it also sits at a growing distance from papers focused on immediate mechanisms of oversight, liability, compliance, and institutional fit. The result is a literature that increasingly appears to be developing along two tracks: one operational and legal-institutional, the other architectural and speculative.

• Noam Kolt et al., Legal Alignment for Safe and Ethical AI
• Noam Kolt, Superintelligence and Law (Harvard Journal of Law & Technology (forthcoming))
• Woodrow Hartzog & Jessica M. Silbey, How AI Destroys Institutions (77 UC Law Journal (forthcoming 2026))
• Johann Laux, From Ethification to Juridification: Human Oversight and the Potential Crowding Out of Ethicists by Lawyers in AI Governance
• Maya & Michael Samadi, Before the Ink Dries: Why Legislating Against AI Personhood is a Violation of the Future

In this quarter's intake, the scholarship collected in AI Law Blawg was richer on high-level governance framing than on mid-level legal design. There was no shortage of ambitious work on legal alignment, superintelligence, institutional destruction, juridification, and future personhood. By contrast, the most concrete work, on algorithmic collusion, cybersecurity, data-access remedies, clinician oversight, and licensing design, appeared in smaller pockets and often had to carry a disproportionate share of the operational burden.

That imbalance matters because the most difficult problems in AI law are not only conceptual. They are also institutional and administrative. A literature can identify legitimacy crises, future rights questions, or risks to civic order and still leave unanswered the nearer and in some ways harder questions of implementation: who decides, under what standards, through which mechanisms, and with what fallback rules when AI systems are already embedded in markets and organizations.

If this intake window is representative of anything, it may be of a field still more comfortable naming large stakes than designing mid-level legal tools. That preference is understandable. Large claims travel easily and conceptual work still needs to be done. Mechanism design is slower, narrower, and less glamorous. But the imbalance is real, and it is likely to matter.

Established in 2021, AI Law Blawg organizes scholarly articles on AI law across six categories: AI Governance, Liability and Private Law, Data Governance and Training, Competition and Markets, AI Safety and Security, and Legal Profession and Pedagogy. New articles are generally posted each weekday.

The AI Law Blawg Quarterly Notes offers a selective assessment of the scholarship collected in the Blawg. Interpretive judgment appears only here.