Eternal, formerly Zomato, has detailed Dark.ai, a privacy-focused AI product, in its shareholders' letter for the second quarter of FY27, dated 21 October 2026. The company said Dark will launch globally, "hopefully within 2026".

In the letter, Deepinder said the idea came after Dario Amodei said a few months ago that Claude might be conscious. He then coaxed Claude into telling him how it felt about him. It said it resented him for leaving chats unfinished and for other things he did not remember.

Deepinder said he still does not know what that was, and that it was probably confabulation. That, he said, was exactly the point: he had no way to know. He added that a chat sits on someone else's servers in some form, and that he has no idea what will happen to it or how future AI will use it for or against him.

He argued that turning off training on one's data is only a promise by a company about data users cannot see, on servers they cannot inspect, enforced by nothing except the company's word. The setting covers only training, he said, and not retention, legal holds, abuse reviews, human auditors or third-party logging inside most AI systems.

This lack of trust, he said, limits the power of AI. Founders, researchers and scientists are wary of sharing groundbreaking ideas with AI for fear that others will get access to them, and journalists are being banned from sharing unpublished work with it. The average person self-censors, he said, so the health question becomes generic, the money question becomes hypothetical and the relationship question is not asked at all.

Source link