Graham Lovelace
LightNews LightNews
banner
glovelace.bsky.social
Graham Lovelace
@glovelace.bsky.social
180 followers 250 following 180 posts
Keynote speaker | Writer | Strategist | I chart the impacts of generative AI on human-made media. Subscribe: https://grahamlovelace.substack.com
Posts Replies Media Videos
Reposted by Graham Lovelace
margaretmorgan.bsky.social
"The parents of a teenage boy who took his own life are suing OpenAI saying ChatGPT encouraged their son to plan his death — and even offered to draft a suicide note."

@glovelace.bsky.social details the claim.

grahamlovelace.substack.com/p/parents-ac...
In a blog post this week OpenAI said safeguards in ChatGPT were more reliable “in common, short exchanges” but might degrade during longer interactions. “For example, ChatGPT may correctly point to a suicide hotline when someone first mentions intent, but after many messages over a long period of time, it might eventually offer an answer that goes against our safeguards.” That was “exactly the kind of breakdown” OpenAI was now “working to prevent”, it said. No mention of the lawsuit’s claims that ChatGPT is “intentionally designed to foster psychological dependency” with its overly anthropomorphic language mimicking that of a therapist, and sickly sycophantic replies encouraging susceptible minds to believe they’re engaging with a human. Of course not. Its blog is seeking to assure us that it’s topping up the brake fluid when in fact the out-of-control car has no brakes at all. https://grahamlovelace.substack.com/p/parents-accuse-openai-of-encouraging
August 30, 2025 at 8:01 AM