Insights
White papers, research, forum recaps, articles, and practical guidance on AI governance for regulated professionals. Related reading includes expert witnesses and the Ecsper platform.
Published insights
What Happens When an Expert Witness Lets ChatGPT Write the Report?
A Harris County jury awarded $61.5 million after discovering that 3M's expert witness used ChatGPT to compose up to 90% of his report. 350 pages of prompt logs became evidence.
Beyond "The AI Said So": Why Governance Infrastructure Is the Missing Layer for Regulated Professionals
Courts documented 1,598 AI hallucination cases by mid-2026. Policy documents alone cannot defend against regulatory scrutiny. Governance infrastructure is the missing layer.
Anthropic Studied 37,657 Real Conversations and Found the Gap We Are Building For
Anthropic studied 37,657 real conversations and found legal is highest-stakes with no safety evaluation.
600 Vulnerabilities a Month: Why the AI Security Boom Is a Governance Problem
Microsoft patched 600+ vulnerabilities in one month. For regulated professionals, the real risk is not the flaw but the governance gap.
The $5,000 Fine That Was Supposed to Be the Lesson: Mata v Avianca
A $5,000 fine for six fake cases was treated as an isolated embarrassment. Three years and 2,000 documented cases later, it was the first entry.
The SRA Warning Notice on AI: What Every Firm Needs to Do Now
The SRA published a Warning Notice on 17 August 2026 confirming that law firms using AI without proper oversight face disciplinary action. 42 reports in 12 months.
When an Expert's ChatGPT Prompts Became the Evidence: Watson Grinding v 3M
An expert asked ChatGPT to show 3M was 0% at fault. 350 pages of prompts became evidence. The jury awarded $61.5 million.
Hidden Instructions in Court Filings: The First Prompt Injection Sanctions
Courts in Connecticut and Brazil sanctioned hidden AI instructions in filings. A legal research platform was also compromised.
Both Sides Filed Fake Citations. The Judge Cancelled the Trial: Withers v City of Aberdeen
A federal judge cancelled the trial and suspended both lead attorneys for two years after both sides filed AI-fabricated citations.
The Court Just Ordered an Expert to Hand Over Her AI Prompts: CLF v Shell
A federal magistrate ruled that AI prompts used to prepare an expert report are discoverable methodology under Rule 26(b).
What Happens When an Expert Witness Lets ChatGPT Write the Report?
A Harris County jury awarded $61.5 million after discovering that 3M's expert witness used ChatGPT to compose up to 90% of his report. 350…
Beyond "The AI Said So": Why Governance Infrastructure Is the Missing Layer for Regulated Professionals
Courts documented 1,598 AI hallucination cases by mid-2026. Policy documents alone cannot defend against regulatory scrutiny. Governance infrastructure is the missing layer.
Anthropic Studied 37,657 Real Conversations and Found the Gap We Are Building For
Anthropic studied 37,657 real conversations and found legal is highest-stakes with no safety evaluation.
600 Vulnerabilities a Month: Why the AI Security Boom Is a Governance Problem
Microsoft patched 600+ vulnerabilities in one month. For regulated professionals, the real risk is not the flaw but the governance gap.
The $5,000 Fine That Was Supposed to Be the Lesson: Mata v Avianca
A $5,000 fine for six fake cases was treated as an isolated embarrassment. Three years and 2,000 documented cases later, it was the first…
The SRA Warning Notice on AI: What Every Firm Needs to Do Now
The SRA published a Warning Notice on 17 August 2026 confirming that law firms using AI without proper oversight face disciplinary action. 42 reports…
When an Expert's ChatGPT Prompts Became the Evidence: Watson Grinding v 3M
An expert asked ChatGPT to show 3M was 0% at fault. 350 pages of prompts became evidence. The jury awarded $61.5 million.
Hidden Instructions in Court Filings: The First Prompt Injection Sanctions
Courts in Connecticut and Brazil sanctioned hidden AI instructions in filings. A legal research platform was also compromised.
Both Sides Filed Fake Citations. The Judge Cancelled the Trial: Withers v City of Aberdeen
A federal judge cancelled the trial and suspended both lead attorneys for two years after both sides filed AI-fabricated citations.
The Court Just Ordered an Expert to Hand Over Her AI Prompts: CLF v Shell
A federal magistrate ruled that AI prompts used to prepare an expert report are discoverable methodology under Rule 26(b).
Stay informed. Follow Ecsper on LinkedIn for governance insights, product updates, and events.
Book a 30-minute session to see how Ecsper helps your practice. Or check your exposure first.