When you ask ChatGPT a question, you're in control of the information you're sharing, EFF’s Lena Cohen told USA Today, but “when you're using the Atlas browser for day-to-day tasks, it's easier to forget that all of your activity could be sent to OpenAI."
https://www.usatoday.com/story/tech/2025/10/22/open-ai-launches-chatgpt-atlas-web-browser/86833766007/
Discussion
Loading...
Discussion
@eff Let me share a story: I'm totally #blind and AI is a precious aid with its image descriptions and immediate information features. We're very vulnerable to privacy intrusion, trying to rely on tech when #accessibility isn't guaranteed and no trusted human is around. Then, "trusted", no one is 100% trusted when it concerns your life. Even the most trusted human could take advantage of you. (1/3)
Tech helps, but is controlled by folks who don't care about vulnerability unless it allows them to earn more money.
For now, when I talk about AI around, I feel very responsible, accompanying enthusiasm to prudence. It has allucination, privacy is not granted... I'm an HIV advocate myself, and have performed the test with photo description using two home-tests for HIV. One negative -mine-, the other positive -my best friend's-. (2/3)
AI reported and described both correctly, I did not go in-deep with questions "could you give me some counseling locations and associations in my area?" It's very risky so the battle must continue both sides: tech MUST be blind-friendly _without_ AI or Internet connection. And AI must be transparent. For now I would not suggest to use AI for real HIV self-test need. (3/3)
"What to know before downloading"
Are you even f*cking serious?
'before downloading'?
Don't
Oh
That's USA Today: pure dreck
bonfire.cafe
A space for Bonfire maintainers and contributors to communicate
Automatic federation enabled