Hacker News Reader

Claiming Arstechnica is jumping the gun is pretty generous to them. They are deliberately lying. OpenAI's default policy is not to share user data without a subpoena. This is standard; every company does this. No reasonable person would position this as "selectively hiding" data. Yet that is exactly how the propagandists at Arstechnica described OpenAI in their headline.

View on HN

> OpenAI's default policy is not to share user data without a subpoena.

As noted in the article, the plaintiffs assert that OpenAI's terms of service state the content belongs to the user, and now it belongs to the user's estate.

So it's not (yet) a question of subpoenas, but about that contract.

Their TOS says the copyright belongs to the user. But I don't see anything in the TOS saying that OpenAI is committed to delivering a copy of the data to the users estate.

True, but it does constrain what justification OpenAI may (credibly) put forward as it navigates in the paired realms of legality and public-relations. AFAIK there's still a lot of temporizing "we are reviewing this" from the company, but that probably can't last forever.

In other words, it makes a difference for OpenAI in deciding between choices such as "we'd love to help but legally can't" or "we could but we won't because we don't want to."

Reply on news.ycombinator.com