A licensing deal sent Character.AI's founders back to Google

The design

On 2 August 2024, Character.AI published a post titled Our Next Phase of Growth, attributed to "The Character Leadership Team." It describes an agreement giving Google a non-exclusive license to Character.AI's current large language model technology, which the post says brings "increased funding" for the company to keep building consumer AI products. Co-founders Noam Shazeer and Daniel De Freitas, with "certain members" of the research team, joined Google; the post says "most of Character.AI's talented team will remain." Dominic Perella, General Counsel and a leadership-team member since mid-2023, became interim CEO, effective immediately. No dollar figure is named for the license or the funding, and the post frames the change as continuity rather than a sale.

What the evidence says

The post is the only primary account of the deal's structure located here; no matching Google statement was found. A later post, announcing Karandeep Anand as CEO in June 2025, corroborates the arrangement without revisiting it directly: Perella held the role roughly ten months before moving to Chief Legal Officer, and Anand is an outside hire, not Shazeer or De Freitas returning. That continuity supports the claim that control did change hands, though Google did not formally acquire the company. "Increased funding" remains a company claim; no document reviewed states an amount, structure, or duration.

What it asks of people

For users, the arrangement asked for nothing directly: no new consent, data change, or price shift appears in the post. What changed is who controls the company's direction, as two consequential engineers and the model lineage they built moved to a different employer while the product kept running under new management. Anand's later post promised faster shipping on memory and content filtering, suggesting the incoming leadership treated the departure as a moment to reset the roadmap, not only to stabilize it.

Privacy and safeguards

Neither post addresses data protection, retention, or how licensing "current LLM technology" affects models already running conversations. A non-exclusive license is narrower than a sale: Google gains rights to use the technology, but Character.AI keeps operating it, and neither post indicates Google gains access to user conversation data through the license itself. That distinction matters against any shorthand headline suggesting Google simply bought Character.AI.

• Did operational authority change hands, or only research staffing?
• What does a non-exclusive license commit the licensee to, and what does it withhold?
• Who sets the product and safety roadmap now, and how has that shown up in later releases?

Character.AI's own account describes a structure built to look like continuity: the founders leave, funding arrives, an interim executive already inside the company steps up, and the product keeps its name. Whether that continuity extended to safety and moderation priorities is separate, and worth tracking against what the company's later leadership actually shipped.