Sentient taps Mozilla open-source AI veteran Julie Belião as it pushes into model provenance

Sentient Foundation has named Julie Belião, former VP of Product at Mozilla.ai and a longtime Google engineer, as its VP of Product.

Belião is now tasked with turning the nonprofit’s research on proving where AI models come from into software developers can actually use.

What is Julie Belião’s role at Sentient Foundation?

Julie Belião, who was previously VP of Product at Mozilla.ai and spent years as an engineer at Google (NASDAQ: GOOGL), has been hired as the Vice President of Product at Sentient Foundation.

Belião’s role is to lead product work in recursive self-improvement (AI that improves itself) and model provenance. The foundation says there is no standard way to prove where an AI model was trained on yet, so Belião’s job is to build working products that can.

During her time at Mozilla.ai, Belião ran the open-source plan for 61 repositories and helped grow its developer community. In 2025, she also co-authored “Towards Best Practices for Open Datasets for LLM Training” with Mozilla and EleutherAI.

The Mozilla Foundation lists her as a collaborator on the paper, along with EleutherAI executive director Stella Biderman and Mozilla’s Mark Surman. The paper argued that AI training data should be licensed, well documented and have a clear origin.

Before Mozilla, Belião spent more than ten years working on multilingual AI. She worked on speech and language data for Google Assistant, Google Translate and YouTube, all products of Google (NASDAQ: GOOGL). She has a PhD in computer science from Université Paris Nanterre.

Why does AI model provenance matter now?

In late September, OpenAI said it stopped a group effort to pull the step-by-step reasoning out of its AI models. It linked the effort to people tied to Moonshot AI, the maker of Kimi.

In February, Anthropic reported that three Chinese AI labs sent more than 16 million requests to its Claude model using about 24,000 fake accounts. Anthropic said the labs wanted to extract Claude’s abilities. Moonshot was responsible for over 3.4 million exchanges with Claude.

When a model is copied, fine-tuned or distilled into another system, buyers often cannot tell what trained it.

Sentient already tried to tackle this problem in October when it launched the Fingerprinting Leaderboard, an open test of whether methods for proving a model’s origin can survive copying, fine-tuning and deliberate attempts to remove them.

Sentient’s researchers built attacks against ten proposed fingerprinting methods and the results showed that nearly all could be beaten without ruining the model, including Sentient’s own Perinucleus method, which had placed 24,576 fingerprints into Llama-3.1-8B.

“Fingerprinting puts the proof inside the model,” Sachi Kamiya, the foundation’s director of venture and growth, explained. She compared it to the hidden marks Renaissance painters used to claim their work.

The research builds on Sentient’s OML work, which is short for Open-access, Monetizable, and Loyal AI Model Serving. It was done with Princeton and the University of Washington and aims to let creators share open models while still proving ownership later.

Sentient launched in February as a Singapore-based nonprofit that promises to keep AGI open-source. In June, it committed $42 million in grants and investments to open-source builders. Its projects, including ROMA, OpenDeepSearch and its OML fingerprinting work, have earned more than 12,500 GitHub stars.

Researchers at Google (NASDAQ: GOOGL), Microsoft (NASDAQ: MSFT) and Meta (NASDAQ: META) have cited them.