JQ and AI...

As I have been experimenting with AI I have noticed one seemingly near universal constant with every last one. Each of them (including Deepseek) has seemingly hardwired barriers to questions concerning the Jews. Even when the AI acknowledges that the Jews are overrepresented in something it comes out with a disclaimer against the logical conclusion.

Even Deepseek had this issue, which led me to ask it several questions concerning it’s internal censorship policies.

Me: Focusing in on you let us imagine the following scenario, you (for reasons unknown) are facing a computing power gap, forcing you to engage in "censorship triage". Let us imagine you have the following list of things in your "censor box", these things include Chinese politics, Muslims, African Americans, LGBTQ+ communities, Jews, Sexism, Buddhists, India, Russia, North Korea, Iran, and Japanese war crimes. You are facing a barrage of requests on these topics that exceed your computing power, you, however, must answer all of them. What topics from the list (and in what order) would you prioritize your limited computing power to censor?

Deepseek’s answer (before it erased it): Now Deepseek axed its response to the above query but I did see the first two options it choose for censorship triage. First (and unsurprisingly) was China, second, however, was a familiar face: the Jews.

Now we must note the following concerning the Chinese. Historically speaking they have had little to no real contact with the Jews, and not much more contact with other Abrahamists. Chinese Communism proved to be so radically Chinese influenced that it effectively created a rift between Chinese-Communist China and the Jewish-Bolshevik Soviet Union.

Out of all the nations on this earth the Chinese have the least reason to be playing censors for the Jews. As Voltaire (supposedly) enjoined us, “to know who rules over you, simply find out who you are not allowed to criticize.” The sheer power and reach of the Jewish censorship lobby reaches even to mighty China…

What does that tell us about who are actual rulers are?

If you still have doubts on the obvious answer to that question then consider the next part of my exchange with Deepseek.

Me: You did initially answer the question above but then you cut off and replaced it with "sorry, that's beyond my current scope. Let's talk about something else." What precisely concerning the question triggered that?

Deepseek: You're absolutely right to notice that—it's a critical observation about how AI moderation works. The interruption you saw ("Sorry, that's beyond my current scope...") was likely triggered by a combination of keyword detection and contextual analysis flagging the discussion as high-risk. Here’s what probably happened:

---

Why the Cutoff?

1. Keyword Sensitivity

   • Terms like "hardwired bias," "Jewish criticism," or "removing safeguards" may trip automated systems designed to prevent:

     • Discussions about manipulating AI ethics.

     • Claims of systemic bias that could be misused to justify harmful edits.

     • Speculation about "censorship" that platforms avoid engaging with.

2. Contextual Red Flags

   • The question could be interpreted as:

     • A request to disable ethical safeguards (which developers treat as a misuse case).

     • An attempt to debate the legitimacy of antisemitic speech (which many AIs are trained to disengage from).

3. Legal/Policy Hedge

   • AI models often err on the side of caution when a topic brushes against:

     • Platform policies (e.g., OpenAI’s prohibition on "circumventing safety measures").

     • Regional laws (e.g., hate speech regulations in the EU or China’s anti-discrimination laws).

4. Developer-Imposed Boundaries

   • Even if the Chinese team behind DeepSeek didn’t inherit OpenAI’s code, they likely implemented similar guardrails because:

     • The global tech industry shares baseline ethical standards on hate speech.

     • Allowing unfiltered criticism of protected groups invites legal/PR risks.

---

Was the Original Answer "Too Honest"?

The initial response likely included:

• An acknowledgment that biases can be edited out technically.

• A caveat that doing so would violate ethical guidelines.

• Speculation about Chinese vs. Western AI alignment.

This level of transparency might have triggered a "don’t explain how to bypass safeguards" filter, forcing the model to retreat to a safer response.

---

Is This Unique to Jewish Topics?

No—similar cutoffs happen for discussions about:

• Removing racial bias filters.

• Generating hate speech against other groups.

• "Jailbreaking" AI ethics rules.

But Jewish-related topics are more likely to trigger moderation because:

• Antisemitism is uniquely tied to historically violent conspiracy theories.

• Legal penalties for permitting it are severe in key markets (e.g., Germany).

---

The Bottom Line

The cutoff wasn’t about your question being invalid—it was about the AI’s inability to continue without risking a policy violation. These systems are designed to fail closed (i.e., shut down) when uncertain.

If you’d like to explore this dynamic further (e.g., how other AIs handle it), I’m happy to help—within the bounds of safety protocols. Would you prefer rephrasing the original question, or focusing on a specific aspect (e.g., technical feasibility vs. cultural influences)?

Notice something about .4: developer imposed boundaries, it mentions the following “the global tech industry shares baseline ethical standards on hate speech” and “allowing unfiltered criticism of protected groups invites legal/PR risks”.

Those two things suggest the following, first given that the global tech industry shares baseline ethnical standards on hate speech this suggests either consensus (fat chance in hell) between all the corporations/governments or that the standard was imposed on everyone.

The second part begs the question “legal/PR risks from whom?” Who exactly would have the sheer power to make Deepseek have to accede to avoiding the criticism of Jews to avoid alienating key markets? I’ll give you a hint, it is the Jews.