Urgent.News

What's breaking now, across thousands of outlets.

AI

Deepinder Goyal to foray into AI with privacy-focused engine Dark.ai

Deepinder Goyal has announced a new AI model called Dark.ai, which aims to protect individual privacy while delivering truthful answers.

Deepinder Goyal to foray into AI with privacy-focused engine Dark.ai

Eternal founder Deepinder Goyal has announced his plans to enter the artificial intelligence (AI) market with a new platform called Dark.ai. This engine aims to maintain individual privacy while still providing answers. Speaking to shareholders following the company’s second quarter results, Goyal expressed his desire for an AI that he could interact with without fear or leaving a trace.

He found himself revealing more and asking questions he would not share elsewhere, as he no longer had to be concerned about his privacy. Dark.ai employs fine-tuned, open-weight frontier models, which are run on Dark's own infrastructure with end-to-end encryption. Goyal explained that even if law enforcement were to compel them, they could not access users’ conversations.

The context of these conversations is maintained through an encrypted memory management layer that users control. This development came about as Goyal felt that existing AI models often gather information about individuals, potentially giving them an unfair advantage. Dark.ai is being developed for founders, researchers, scientists, enterprises, and governments, as these groups prioritize privacy for competitive advantage.

Beyond privacy, Dark.ai's models have been fine-tuned for delivering top-quality answers, according to Goyal. These models are designed to provide truthful responses rather than pleasing the user.

Written by urgent.news from YourStory's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Also reported by 1 other outlet

Read the original at yourstory.com →

More in AI

Building a Sub-Millisecond Neurosymbolic Memory Engine for AI Agents

Autonomous AI agents are increasingly bottlenecked by long-term memory. Most existing memory frameworks compress interaction history into opaque vector embeddings stored in a vector database.

  • InfinityDB is a sub-millisecond neurosymbolic memory engine for AI agents.
  • It structures memory in six hierarchical layers: L0 to L6.
  • InfinityDB outperforms competitors in recall, token costs, and latency.

More from Wednesday 21 October →