Perhaps the most important part of evaluating a software dependency is identifying it's structural risks to your business/project. Eg, if you're relying on a business's product, does them going bust mean the end for your project.

In Google's case, they often will create an open source project in license, but not community. They take the reigns and hold them so tightly that if they let go, nobody is there to pick them up. Compare this to projects created by a single individual: for those to grow in contributors, diversification happens automatically. Transitioning a project in a healthy way requires others to already be actively contributing, and probably in a minor leadership role.

The thing here isn't that Google is obligated to spend engineering resources on this forever. It's that the systemic risk of choosing a major Google product at first appears low because Google is one of the most successful companies in existence. However, it's actually very high because they have a history of the above behavior.

Legally obligated, no. But there is an obligation of a certain level of reliability in order for a party to maintain a reputation that their products are worth adopting for anything important.