Dear Opera Support Team,
Subject:1. Repeated Overblocking of Legal Case Commentary of an Actual Trial Record and 2. the Repeated Loss of Lengthy, Long-term AI Conversations since October 2025
**
I am a user of the Opera AI assistant and have been working on a complex, long-term legal writing project involving detailed case analysis. I have faced challenges with session continuity and content guidelines that impact my workflow.
Below, I outline these two issues and how they impact my work. I hope my feedback helps improve the assistant’s functionality for users engaged in extended, in-depth, legal research projects. My letter to Support/Developers is below. I have broken it into two parts, A + B, to meet your charcter length requirements
Hello Support/Developers,
Part B: I’m asking you for your help with two issues that make it impractical for me to use the assistant for a long‑form, law-and-trial-based, book project I've undertaken.
The book is about the Trial and Conviction of OCA Archbishop Seraphim For An Imaginary Crime. This was a trial reported Internationally. I reference Canada court transcripts directly from the preliminary evidential hearing through to the trial and later to Sentencing. None of those transcripts, nor the ruling, nor the sentencing have been published online or anywhere. My book closely, carefully discusses how the viva voce evidence emerged from failed or unintentionally falsified historical memories which led to the wrongful conviction of that Orthodox prelate, the Archbishop for the OCA Archdiocese of Canada, for a crime that never took place.
I repeatedly must cite, refer to, and analyze testimony, timelines, and two wrong chronologies, all of which from the evidential basis for the case but which are based solely on fragile and faded memories 'recalled' decades after the imagined crimes were alleged to have happened. Save for the accused and a key prosecution witness {each public figures}, I've disguised all testifiers and their locations wherever needed to protect their privacy or comply with court records.
Let me continue this letter in Part B below. You should receive it within the next 10 to 15 minutes.
The Issues:
Issue 1 is the overblocking of lawful case references preventing useful AI assistance.
Issue 2 is how my various conversations with your AI assistant vanish even when I have not exited from nor sought to close them..
Re Issue 1: Whenever I describe the allegations (the “imagined crime”) or select oral testimonial evidence to analyze credibility, reliability and chronology—the assistant sometimes refuses to engage, stating this violates guidelines. This is triggered by words such as sexual assault, naked, and similar terms or expressions of a sexual nature which, given the nature of the alleged crime (sexual assault), necessarily prevents by fiat any and all substantive feedback on the core subject matter of the book (real trial, public courtroom proceedings, with an analytical critique of the evidential basis of the trial that finds major viva voce evidence from memory provably unsupported, incredible or unreliable.
I sought help only where the evidential basis (testimony) was provably neither credible nor reliable (and always unsupported by other than the complainants or the key witness for the prosecution. I intent to show where their inaccuracies badly steered every judicial assessment of the credibility and reliability of witnesses, prosecution or defence.
Yours sincerely,
Larry Motuz
