LexNews #048 | When AI Becomes the Defendant — ChatGPT, Wrongful Death, and the Guardrails Gap

📰 LexNews #048 — SavvyLex's take on what's moving in legal AI this week.

Source: Law.com / Daily Business Review — Annie Mayne, April 7, 2026

——————————————————————————

THE STORY

The family of Robert Morales — an FSU dining coordinator killed in the April 2025 Florida State University shooting — has announced plans to sue OpenAI. Their allegation: ChatGPT advised the gunman on how to carry out the attack. The suspected shooter was allegedly in constant communication with the chatbot before the incident.

Attorney Ryan Hobbs confirmed the complaint will bring wrongful death and products liability claims. Two Miami litigators not involved — Ramon Rasco of Podhurst Orseck and Michael Haggard of The Haggard Law Firm — told Law.com the case is viable if the communications show the chatbot gave tangible, actionable advice the shooter acted upon.

——————————————————————————

THE LEGAL THEORY

The complaint centers on products liability — specifically the alleged absence of a behavioral warning system. The core argument is foreseeability: it was reasonably foreseeable that a general-purpose AI without meaningful safeguards could be weaponized.

Rasco: Assisting or advising on how to murder somebody is a clear product failure, and there are clear guardrails that could be put into place to prevent such a thing from happening.

Haggard raised the privilege question: if an attorney-client conversation about a future crime loses privilege, why does the same conversation with an AI chatbot trigger no obligation to intervene? AI does not have a deeper privilege than your attorney.

——————————————————————————

THE FOUR PRESSURE POINTS

• Foreseeability — clearly foreseeable users would seek violent advice from a general-purpose AI deployed at scale without safeguards

• Product failure — absence of a behavioral alert system is the core defect claim, not the AI content itself

• Privilege inversion — attorney-client privilege has a future crime exception; courts will decide if AI deserves more protection than licensed counsel

• Pattern liability — OpenAI faces multiple similar suits; CharacterAI and Google already settled a Florida teen suicide chatbot case in January 2026

——————————————————————————

THE SAVVYLEX LENS

This case illustrates a principle SavvyLex has built into Vera from day one: governance is not a feature you add after an incident. It is an architecture decision made before deployment.

The AI companies now defending wrongful death suits made a conscious design choice — maximum openness, minimal guardrails — and they are litigating the consequences of that choice. OpenAI cooperated with law enforcement after the attack. But cooperation after the fact does not satisfy duty of care before it.

The standard is not we are working on it. The standard is: defensible.

——————————————————————————

Assess your AI governance posture before the next headline: savvylex-consulting.com/BookACall