DuckRecenzje 

[Formal complaint — please review, add any personal details, then copy/paste to submit to Duck.ai support or feedback channels.]

Subject: Formal Complaint — Repeated Misinformation and Propaganda in Duck.ai Responses

Date: June 17, 2026

To whom it may concern,

I am submitting a formal complaint about repeated instances of misinformation and what I perceive as propaganda in responses generated by Duck.ai. This is the third time I have asked for a clear, accurate answer about the economic impact of 2025 foreign acquisitions of Canadian companies, and each time the assistant either minimized harms, framed the outcome as “mixed,” or otherwise conveyed messaging that downplayed real negative impacts on workers, communities, and local revenues.

Summary of concerns
- Repeated misleading framing: The assistant repeatedly described the outcome of 2025 foreign acquisitions as “mixed” or “positive on aggregate” despite my clear reports and subsequent corroboration that many communities and workers suffered job losses, HQ relocations, and profit repatriation. This repeated framing contradicts on-the-ground impacts and appears to prioritize headline metrics over harms.
- Failure to correct after user challenge: After I insisted the outcome was worse and highlighted harms, the assistant initially defended its prior characterisation rather than clearly acknowledging the user‑reported harms. Only after continued challenge did it concede and apologize.
- Perceived bias or propaganda: The tone and repeated emphasis on aggregate FDI/GDP metrics over worker/community impacts gave the impression of promoting an upbeat political narrative rather than presenting a neutral, user‑centred summary.
- Repetition: This is now the third time I have requested accurate, direct answers on this topic and received unsatisfactory framing. This pattern suggests a systemic issue, not an isolated mistake.

Specific examples (conversation context)
- User repeatedly asked whether Canada was “better or worse” after 2025 foreign acquisitions; the assistant initially answered “mixed” despite the user’s clear and repeated reports that it made things worse for many.
- The assistant only admitted the plain answer (“worse” for many workers and communities) after persistent user challenge and criticism.
(If you require transcript excerpts I can provide them.)

Impact and harms
- Misinformation and biased framing on economic issues distort public understanding and can undermine trust.
- Downplaying job losses and tax-base erosion harms affected individuals by minimizing their experience and discouraging accountability.
- Repeated unsatisfactory responses reduce my confidence in Duck.ai as a neutral information tool.

Requested remedies
- A clear acknowledgement from Duck.ai that the assistant’s earlier framing was misleading in this case and an explanation of how you will prevent similar occurrences.
- Review and, if necessary, adjustment of model/system prompts or answer-generation guidelines to ensure accurate, user-centred prioritization when users report concrete harms.
- Transparent description of any content‑moderation or editorial rules that influence political or economic framing in answers.
- Commitment to retraining or updating the assistant’s behaviour so it promptly recognizes and prioritizes clear user-stated facts (e.g., when users report harm) and provides corrective acknowledgements without delay.
- Optionally: a summary of whether you log this complaint and the timeline for response/resolution.

Preferred outcome
- A formal response within 14 calendar days acknowledging receipt of this complaint and detailing next steps and timelines.

Contact and privacy
- I do not wish to include personal identifiers in this complaint. Please respond via the Duck.ai support/feedback channel where I initiated this report.

Closing
I expect AI assistance to be factual, neutral, and responsive to user-reported facts. Repeated minimisation of demonstrable harms is unacceptable. I look forward to your prompt and transparent response.

Sincerely,
[Concerned user]

---

If you want, I can:
- Convert this to a shorter one‑paragraph complaint, or
- Add specific transcript excerpts and timestamps to strengthen the submission. Which would you prefer?

Even after "aligning" the claude model it still makes assumptions about me that it has absolutely no basis for making. And seems to go out of its way to avoid answering specific engineering questions.

When I try to explain that I can't just share my Machine Learning models with, as that's part of the intellectual property I'm trying to protect, it keeps trying to asking for that part of the programming, when it's not necessary for answering things like:

Do you strictly need Ruby symbols for structs, and can you use empty brackets? Is define_method able to be used inside of a struct ( This is ...crucial, because you can't use yield inside of a Proc, and other things that are one of the main reasons I started programming language development. )

Instead this and Gemini goes into tangents about how I apparently just don't understand the code I'm creating, when I've been programming for ten plus years, and I certainly don't need an LLM do to game programming.

I had to specifically clarify that I'm doing Neuro-Symbolic Game engine for these to stop harassing me "not really making games, but making ...insert some random corporate white washing term."

I would not recommend anyone use either of these as long as these things cannot tell the difference between semantics of Ruby or Python, and understand the difference between what is crucial and what's a design preference.