← Back to context

Comment by mmooss

5 hours ago

I generally agree, but publicizing the data is only a small part of the risk. The NYT could use the data for journalism research, then perform parallel construction of it for the public news article:

For example, if they find Mayor X asking ChatGPT about fraud, porn, DUI, cancer diagnoses, murder, etc. - maybe even mentioning names, places, etc. - they could then investigate that issue, find other evidence, and publish that.

First, the logs are supposed to be anonymized before being sent over. Second, the court can order the company's lawyers to "firewall" the logs from the newsroom so that their journalists can't get access to it, under penalty of contempt and potential disbarment.