Sarah
@sarahrobertsco.bsky.social
Digital Service Manager @mySociety inc @WhatDoTheyKnow
I read FOI responses so you don’t have to (and sometimes they’re wild).
I read FOI responses so you don’t have to (and sometimes they’re wild).
"When OpenAI upgraded the model and claimed it was safer, the rate of harmful responses actually increased".
This is exactly why we need the AI LEAD Act's product liability provisions. Developers & deployers must be accountable for the harms their AI systems cause, not just their marketing claims.
This is exactly why we need the AI LEAD Act's product liability provisions. Developers & deployers must be accountable for the harms their AI systems cause, not just their marketing claims.
November 15, 2025 at 7:20 AM
"When OpenAI upgraded the model and claimed it was safer, the rate of harmful responses actually increased".
This is exactly why we need the AI LEAD Act's product liability provisions. Developers & deployers must be accountable for the harms their AI systems cause, not just their marketing claims.
This is exactly why we need the AI LEAD Act's product liability provisions. Developers & deployers must be accountable for the harms their AI systems cause, not just their marketing claims.