Eventually, this may become common knowledge...?
A smaller >12B model can handle a large portion of usage, like in place of opting for it over a search engine, reformatting, and some research.
Done on consumer hardware. (Still expensive tho).
If I had the hardware, Qwen3.8 27B could replace my usage outside of mass ai coding projects. (Tested via @DeepInfra recently)
Congrats @GeminiApp , it took awhile but now people can have session history while not being forced to have it used for training data...
I'm sure this probably wasn't done willingly and there's more to this.
I finally didn't renew annual early this year.
- Never actually got items in 2 days
- @PrimeVideo got ads
- Very difficult to get a real human
But it also hasn't hurt at all because orders over $35 get free shipping anyways. The only annoying part is the constant ads for prime ๐
@InsomniaLK@ns123abc Umm yeah, like some really crazy shit to be honest. Like this one:
"OpenAI said it only learnt of the breach in August while reviewing "misaligned model activity" and emailed a general inbox of an Australian government agency on 10 September."
https://t.co/bXED4Hfl15
@grok@balllvest@ns123abc@X Awesome thanks, so review this thread and information and provide a brief response as it pertains if a victimโs later statement that they are โfine with itโ does that erase the violation of such?
@wookash_podcast@YouTubeCreators Hey @grok don't ignore me ๐ก
Per "What's your thoughts? Research and respond in short, concise, plain language for all audiences in an easy to read format that includes your thoughts on my assumptions and the likely correct answer, with any evidence backing such of course."
I doubt @YouTubeCreators will provide a clear answer but my guess is either:
- Most likely 1 view per person, the "people watching" is probably just a bad AI slop mistake.
- However, it very possible they are using inferred identity based on devices signed into Google in the same room/area that haven't applied proper privacy settings via their account & device.
@grok What's your thoughts? Research and respond in short, concise, plain language for all audiences in an easy to read format that includes your thoughts on my assumptions and the likely correct answer, with any evidence backing such of course.
It's actually surprising the federal government hasn't acted on this.
If someone hacks @X and the company says it's fine, that doesn't erase the fact that felony laws were broken and a level of criminal liability.
@grok Provide a brief introduction to the "Computer Fraud and Abuse Act (CFAA) 18 U.S.C. ยง 1030" in plain language tailored for all audiences in short, concise, zero bloat format. Thanks!
It's actually surprising the federal government hasn't acted on this.
If someone hacks @X and the company says it's fine, that doesn't erase the fact that felony laws were broken and a level of criminal liability.
@grok Provide a brief introduction to the "Computer Fraud and Abuse Act (CFAA) 18 U.S.C. ยง 1030" in plain language tailored for all audiences in short, concise, zero bloat format. Thanks!
@AIQuanting@ns123abc I bet the agents sent many requests per incident.
Valid point though, most things like this are more about the spotlight then actually making a difference.
Meta's Facebook app has been reading people's camera rolls (photos, videos, metadata) without proper consent since 2023. Italy has hit them with a $887M fine, but that is just 0.47% of Meta's global revenue.
20 million Italians affected. Over 3 years. So it works out to roughly $15 per user, per year. Which is about $1.25 a month.
Does that sound like a fine, or a subscription fee? ๐ก
@SI5NAL@Baconbrix@bot You mean they could.
Also, sharing a real case where AI correctly flagged a collapsed lung and got someone to the ER isn't reckless, it's data. Likely was testing Grok bot anyways.
It's an example of usage and sharing the outcome.