Mark Suman, co-founder and CEO of Maple AI, used an interview with BMTV to outline a private AI model designed so that even Maple's own team cannot read users' chats.
Suman said AI is becoming embedded in more aspects of life, including financial transactions and agents paying each other. In that context, he said he would not want such chats read by a company whose source code cannot be inspected.
Built on open-source code and encryption
Maple, according to Suman, was built from the ground up with open-source code and encryption. He compared the approach to Bitcoin's emphasis on verification: rather than asking users to trust the company, he wants them to be able to verify.
Suman said user data, thoughts and other sensitive information remain encrypted throughout the pipeline. He also said users do not have to sacrifice security to use the best open-weight models.
Topics covered in the conversation
The BMTV discussion also touched on scenarios including a warrant arriving at Maple, advertising steering AI intelligence, risks to user data, agents approving their own budgets, and hard limits on what an agent can do. It addressed paying for private AI with Bitcoin and what might push everyday people toward private AI.
Suman's broader argument centered on inspectability and encryption rather than trusting a provider whose code cannot be examined.
