Eternal has unveiled Dark.ai, a privacy-focused artificial intelligence platform that founder Deepinder Goyal aims to launch globally by the end of 2026. The platform is designed for people and organisations that want to use AI for sensitive conversations, confidential research and commercially valuable work without compromising privacy.
According to Goyal’s comments in Eternal’s Q2 FY27 shareholder communication, Dark.ai has already been used internally for several months. The platform will use fine-tuned open-weight AI models running on Eternal’s own infrastructure, alongside end-to-end encryption and an encrypted memory system that users control. Its planned audience includes founders, researchers, scientists, enterprises and governments.
Key Takeaways
- New AI platform: Eternal has introduced Dark.ai, a service built around private AI interactions.
- Global launch target: The company aims to launch the platform globally by the end of 2026.
- Privacy-focused design: Dark.ai is intended to protect conversations and retained context through encryption.
- Open-weight models: The platform will use fine-tuned models running on Eternal’s infrastructure.
- User-controlled memory: An encrypted memory layer is designed to preserve context across sessions under the user’s control.
- Target customers: Founders, researchers, scientists, enterprises and government organisations.
- Key question: Independent verification of the platform’s security architecture and privacy claims will be important as the product becomes available.
What Is Eternal’s Dark.ai Platform?
Dark.ai is Eternal’s entry into privacy-focused AI services, an area attracting interest as individuals and businesses increasingly use AI assistants for research, planning, writing and problem-solving.
Many AI interactions involve information users may not want to expose unnecessarily. Founders could discuss confidential business strategies, researchers might work with unpublished findings, and companies may use AI to analyse sensitive internal documents.
Goyal has argued that concerns about who can access conversations can discourage people from sharing information freely with AI systems. Dark.ai is intended to address that concern by making privacy a central part of the product’s design rather than treating it as an optional setting.
The platform is being developed under Eternal, the parent company of Zomato and Blinkit. However, its intended use differs from the group’s established food-delivery and quick-commerce businesses.
Its commercial positioning will depend on whether Eternal can turn its privacy-focused approach into a product that users trust and find useful in everyday work.
How Dark.ai Plans to Protect User Privacy
Dark.ai’s proposed architecture combines open-weight AI models, infrastructure operated by Eternal and encryption features intended to protect user conversations.
1. End-to-end encryption
Eternal says Dark.ai will use end-to-end encryption to protect conversations. In a properly implemented end-to-end encrypted system, only the intended endpoints possess the keys needed to read the protected content.
Goyal has said that Eternal would not be able to read users’ conversations, even if compelled to do so. This is a significant privacy claim, but the practical protection will depend on how encryption keys are generated, stored and managed, and how the platform handles model processing.
The company has not yet publicly established through independent technical verification that every part of the proposed system meets this standard. Details about key management, security audits and the handling of prompts during AI inference will therefore matter.
2. Encrypted memory across sessions
AI assistants often use previous conversations or stored information to maintain context. This can make an assistant more useful, but it also creates questions about what information is retained and who can access it.
Dark.ai plans to include an encrypted memory management layer that allows context to carry across sessions under the user’s control.
If implemented effectively, this could help users continue complex research or projects without repeatedly explaining their requirements. At the same time, users will need clear controls over what is stored, how long it is retained and how it can be deleted.
3. Models running on Eternal’s infrastructure
Dark.ai will use fine-tuned open-weight models running on Eternal’s own infrastructure.
Open-weight models make trained model parameters available for developers to use and adapt under the relevant licence terms. Fine-tuning allows a company to adjust an existing model for particular tasks or behaviour without necessarily training an entirely new foundation model.
Running models on its own infrastructure could give Eternal greater control over deployment and data handling. However, operating the infrastructure itself does not automatically guarantee privacy. Security also depends on system configuration, access controls, logging practices and the design of the inference pipeline.
Why Dark.ai Is Targeting Founders, Researchers and Enterprises
Eternal is targeting users for whom confidentiality can be especially important.
| Target audience | Potential use cases |
|---|---|
| Founders and entrepreneurs | Business planning, strategy development and confidential brainstorming |
| Researchers and scientists | Exploring hypotheses, analysing research and working with unpublished ideas |
| Enterprises | Internal knowledge work and assistance with sensitive business information |
| Government organisations | Research and administrative tasks involving protected information |
| Individual power users | Private conversations, personal research and complex projects |
For these groups, privacy can influence whether AI is adopted at all. An organisation may hesitate to use a third-party assistant if it cannot establish how sensitive information is stored, processed or accessed.
Dark.ai could appeal to customers who want AI assistance but need stronger guarantees around confidentiality. However, enterprise and government customers typically require detailed security documentation, contractual protections and compliance evidence before approving sensitive workloads.
The platform will therefore need to demonstrate not just useful AI performance but also dependable security and administrative controls.
Dark.ai Enters a Competitive AI Market
Privacy is becoming an important differentiator as AI assistants move into more sensitive professional tasks.
Major AI providers offer different combinations of privacy settings, data-retention controls, encryption and enterprise protections. Some open-weight AI deployments also allow organisations to run models in environments they control.
Dark.ai’s proposed distinction is its emphasis on privacy across both conversations and persistent memory. Rather than relying only on promises about how customer data will be used, the platform aims to make the underlying system architecture central to its privacy proposition.
That positioning could attract users who want to share more context with an AI assistant. But competition will be demanding: customers will compare output quality, speed, reliability, cost, integrations and security, not privacy alone.
Eternal will also need to explain which models power the service, how frequently they are updated, whether users can choose different models and how the platform will handle misuse or security incidents.
What Dark.ai Could Mean for Eternal
The launch would represent a technology initiative beyond Eternal’s established consumer internet operations. Its existing businesses provide experience in running large digital platforms, but building an AI product introduces different technical, competitive and commercial requirements.
A successful launch could give Eternal exposure to the growing market for AI-assisted research and professional productivity. A privacy-led proposition may also help differentiate the platform in a market where customers are increasingly conscious of how their information is handled.
However, Eternal has not disclosed Dark.ai’s pricing, revenue model, investment budget or expected contribution to its financial performance. It is too early to estimate the platform’s commercial impact or determine whether it will become a material business for the group.
The planned global launch by the end of 2026 is a target rather than confirmation that the service will be available in every market on that date.
The Bigger Picture
Dark.ai reflects a wider shift in the AI industry: as assistants become more capable, users increasingly want to share richer context without losing control over their information. Privacy is becoming a product-design issue, not merely a setting in an account menu.
For Eternal, the opportunity is to combine useful AI capabilities with a system that can credibly protect sensitive conversations. The challenge will be demonstrating that the platform’s encryption, memory controls and infrastructure deliver the protection promised. Independent security assessments, clear documentation and transparent data policies will be essential to building trust.
FAQs
1. What is Eternal’s Dark.ai?
Dark.ai is a privacy-focused AI platform announced by Eternal founder Deepinder Goyal. It is designed for users who want AI assistance for confidential conversations, research and professional work.
2. When will Dark.ai launch globally?
Eternal is targeting a global launch by the end of 2026. The company has not confirmed that the platform will be available in every country on a specific date.
3. How will Dark.ai protect user data?
Eternal says the platform will use end-to-end encryption, an encrypted memory management layer and fine-tuned open-weight models running on its own infrastructure. The complete security implementation has not yet been independently verified publicly.
4. Who will be able to use Dark.ai?
The intended audience includes founders, researchers, scientists, enterprises, governments and other users who require greater confidentiality when using AI. Pricing, access requirements and the final availability arrangements have not yet been disclosed.
Looking Ahead
The next milestone for Dark.ai will be a fuller product announcement explaining how users can access the service, which AI models it supports and what privacy controls will be available. Independent technical assessments and clear information about encryption, memory retention and data handling will help users evaluate its claims before relying on it for sensitive work.
Over the longer term, Dark.ai’s success will depend on balancing privacy, model performance, usability and cost. If Eternal can demonstrate that its security architecture works as promised while delivering useful AI capabilities, the platform could find a place among users seeking confidential AI assistance. Until those details are available, its planned launch is an important development to watch rather than proof of an established competitive advantage.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
