New AI tool shames legal system

Andrew L. Urban

He was accused, charged, tried, convicted and sentenced; his appeal was dismissed. His Petition for Mercy to the Governor was put aside by the A-G as was his claim of innocence. His otherwise exemplary life, his family and business were all destroyed. But there has always been a concern that a miscarriage of justice has occurred. Now a new, specialised AI tool called LIA has confirmed it. What is LIA and why should the law act on its findings?

We tasked LIA to investigate this case as a test of LIA’s prowess, based on documents we provided, from the judge’s summing up to the summary of the appeal and the 76 page Petition intended for the Governor seeking a Royal pardon, a last resort that was dismissed by the then Attorney-General.

LIA found multiple significant inconsistencies, including documented factual impossibilities, incompatible third-party witness evidence and new avenues of expert analysis not previously explored. The LIA report shows a litigation-ready pathway to a judicial review and includes the completed application to the Attorney-General. All it needs is a signature. All in just a few minutes.

Adrian Bertino-Clarke

LIA stands for Legal Intel AI, which is currently in advanced live field trials. It takes AI to the next level in several specific areas, eg the criminal law: it’s “a system designed not to replace lawyers, but to detect injustice, construct appeal pathways, and restore fairness where systems have failed,” says Adrian Bertino-Clarke, President & CEO, FIA Labs / Federated Intel AI LLC, an Australian-owned Washington-based research and AI architecture company.

‘Federated’ in this context refers to a federated system of expert AI agents, each trained for a specific high-stakes domain — medicine, law, finance, organizational design, public safety, branding, and ethics.

Bertino-Clarke explains that what “FIA and LIA represent is not the promise of artificial intelligence in the abstract — but of accountable, structured, and legally defensible intelligence. Every step is auditable. Every assumption is engineered. Every output can be tested in court, audit, or oversight review. This isn’t automation. It’s patented architecture.”

Our ‘test drive’ of LIA using a case (historical sexual abuse of a minor) we first reported in 2017 in the firm belief that it had been a wrongful conviction, resulted in the legally founded confirmation that the verdict is unsafe. The LIA report included the following:

5 Legal Standards Relevant to the Enquiry
5.1 The “Unease” Standard (White v The King (1906))
An enquiry may be ordered where material causes genuine “unease” about allowing the conviction to stand.

5.2 Miscarriage Principles (M v The Queen (1994))
A conviction may be unsafe where evidence contains inconsistencies that could raise a reasonable doubt.

5.3 Long Delay and Memory Reliability (Longman v The Queen (1989))
Significant caution is required where allegations involve decades-old memory and no physical evidence.

5.4 Inconsistent or Impossible Evidence (Domican v The Queen; Jones v R)
Material contradictions on central facts can render a conviction unsafe.

5.5 Relevance to s 77—Varley v AG (NSW)
Petitions should be interpreted beneficially; the Executive may act even in the absence of traditionally “fresh” evidence.

7 Issues Potentially Capable of Amounting to a Miscarriage of Justice
7.1 Structural impossibility in complainant’s account
If renovation timelines confirm the physical layout described did not exist, this is fundamentally exculpatory.

7.2 Demonstrably false corroboration
If DD could not have been present in 1986, the corroboration evidence collapses.

7.3 Evolution and expansion of allegations
Progressive embellishment may indicate memory distortion.

7.4 Inadequate testing of evidence at trial
Failure to challenge central inconsistencies may have deprived the jury of critical material.

7.5 Appeal process not addressing critical issues
The CCA did not evaluate the factual reliability problems now raised.

8 Conclusion and Recommendation
Based on:
-the multiple significant inconsistencies;
-documented factual impossibilities;
-incompatible third-party witness evidence;
-new avenues of expert analysis not previously explored;
…there is a properly grounded concern that a miscarriage of justice may have occurred.

Recommendation:
That a judicial officer be appointed under s 77(1)(a) to enquire into the conviction, with full powers to obtain evidence, examine witnesses, and determine whether the conviction should be referred to the Court of Criminal Appeal or whether further action should be taken.

It is an objective, independent and thorough investigation that condemns our legal system. How did a case so riddled with errors and avoidable mistakes pass through every stage uncorrected?

*** Special Access for Wrongful Convictions Report Readers ***

To support the advocacy community, FIA Labs is offering free access to LIA Lite. Visit FIA-Labs.com/LIA and enter promo code CHRISTMAS-2025 at checkout to activate your account.

This Lite version includes core features already in use by Innocence Project teams — designed to help dissect briefs, surface evidentiary concerns, and map viable legal arguments. No credit card required. The standard price is USD $49 per app LIFETIME licence (No subscriptions).

 

 

This entry was posted in General articles. Bookmark the permalink.

One Response to New AI tool shames legal system

  1. Public Acknowledgment:

    We extend our sincere thanks to Wrongful Convictions Report and journalist Andrew L. Urban for the thoughtful and powerful article on LIA — Legal Intel AI. In a legal landscape where complex systems are rarely accessible to those without means, this kind of informed exposure is critical.

    LIA was designed to detect injustice, construct appeal pathways, and restore fairness where systems have failed. Just as importantly, it helps minimise wrongful prosecutions by analysing cases from the evidence up — not through the lens of prosecution or defence, but through evidentiary coherence and factual scrutiny. This structural neutrality ensures fairer, more accountable outcomes.

    LIA’s strength lies not just in its architecture, but in its use. The more people interact with it, the more capable it becomes at solving complex, real-world legal problems. That’s why Andrew’s article is so valuable: it reaches a community deeply committed to justice and uniquely positioned to help sharpen and refine the system.

    We are proud to support this community by offering free LIFETIME access to LIA Lite. Our thanks again to Andrew and WCR for helping advance this mission.
    Federated Intel Ai / FIA Labs

Leave a Reply to Adrian Bertino-Clarke Cancel reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.