From the Rav’s Desk: What Bracha should be said on Dragon Fruit & Why I had to revoke “Semicha” from AI chat boxes [ChatGPT, Microsoft Copilot, Google Gemini, xAI Grok]

Proper Bracha on Dragon fruit (i.e. Pataya)

Question:

What Bracha should be said on Dragon Fruit? I heard different things mentioned in your name with some saying that you said that the blessing is Ha’adama and others saying that you said the blessing is Haeitz. I even asked my AI chat box and it quoted you from your website to say that it is Ha’dama and quoted an entire explanation you gave for why it is Ha’adama. But then I looked at your website and saw that you actually wrote that it is Haeitz, and no such explanation was ever written there. However, in my independent research using AI chat box I discovered that in truth many of the leading Poskim rule that it is Ha’adama, including Rav Elyashiv, Rav Shlomo Zalman Aurbach, and other leading figures. Can you please clarify this matter.

Answer:

My original and current answer remains the same, that the proper blessing to be said over Dragon fruit is Haeitz. This is not to be confused with passionfruit or Passiflora, whose blessing is Ha’adama.

Explanation:

Dragon fruit, known in Hebrew as פִּיטַיָּה (pitaya), grows on a perennial climbing cactus with thick, three-sided stems, producing large bright pink or yellow fruit along the stem edges. Therefore, this fruit is considered equivalent to any other tree fruit for which the blessing of Haeitz is recited. It bears similarity to the cactus fruit known as Sabras, over which the same blessing is made, as explicitly stated in the Poskim. In addition to reaching my own independent conclusion, I have observed that every Bracha Luach I have consulted, as well as every publication addressing the blessing for Dragon fruit, consistently present the same view. There does not appear to be any substantial discourse proposing an alternative blessing, nor am I aware of any halachic authorities who hold a differing opinion. As far as the sources from AI, it is seemingly no more than an AI hallucination, a common issue with current AI chat interfaces that I will discuss below. However, it is noteworthy that there is some debate concerning the laws of Arla as they pertain to dragon fruit, specifically due to that it grows mainly from the air rather than the ground which thereby may make it not be subject to the Arla prohibition. Practically, the Poskim have concluded that the prohibition of Arla applies to dragon fruit as it does to other fruits.

The Accuracy of AI chat boxes when it comes to Shaalos:

It is essential to understand that AI should never be relied upon for halachic rulings. The reason is simple: AI systems often produce inaccurate information and, at times, fabricate sources entirely, which is coined as AI Hallucinations. These hallucinations can be extremely misleading because they may include not just false answers, but even fabricated quotations with detailed references from the greatest of Poskim which make it appear authentic. Only a qualified scholar who carefully verifies the sources that Ai has provided will uncover that the material was entirely fabricated. This is not theoretical—I have personally experienced this issue multiple times with AI, including with the very question of the blessing for dragon fruit. In that case, the AI chatbot quoted my own website incorrectly, claiming that the blessing is Ha’adama when, in fact, I had written Ha’eitz. It even invented an explanation that I so call wrote for why it should be Ha’adama, something I never wrote and fundamentally disagree with. Let this serve as a clear warning: do not rely on AI for halachic questions. Always seek guidance from a competent rabbi. Furthermore, we must stop referring to AI chatbots with titles like “rabbi,” as doing so misrepresents their nature and authority.

The following is a quote from my own Copilot AI which explains the above notion: AI hallucinations are mistakes where an AI generates information that sounds plausible but is actually false or fabricated. They happen because the system predicts patterns in language rather than accessing verified facts, so it may invent details, sources, or events. This is especially risky in areas needing precision, since the AI can present errors with confidence, making them seem trustworthy. Accordingly, AI sometimes fabricates halachic sources because it generates text by predicting patterns rather than accessing true knowledge, and since halachic literature is vast and specialized, gaps in its training data lead it to produce plausible but incorrect citations; this is especially problematic in halacha, where precision and authority are critical, so a misattributed ruling can be misleading or harmful, which is why the safest approach is to treat AI answers as starting points, verify them against reliable databases like Bar Ilan Responsa or Sefaria, and rely on AI more for plain-language explanations than for exact citations.”

Sources: See regarding Arla by dragon fruit: Techumim 26 p. 57-67; See regarding Sabra – Cactus Fruit: Admur 203:1; Birkeiy Yosef 202:1; Kaf Hachaim 203:6; Ketzos Hashulchan 49:19; Kerem Tziyon 6:8; Shevet Halevi 4:23; Yechaveh Daas 2:21; Piskeiy Teshuvos 203:4

About The Author