:marseywall: “ Well Meta's 65 billion parameter language model just got leaked to the public internet, that was fast. Get ready for loads of personalized spam and phishing attempts. Open sourcing these models was a terrible idea”

https://x.com/jeffladish/status/1631825647436980224
90
Jump in the discussion.

No email address required.

Lol what will gmail do now? Instead of a billion spam emails getting auto-filtered they’ll have a billion * 0.01% by the r-slurs who think spending $300k - $1M+ training an LLM will actually help them (the leaked params are not the same as leaking a full model with an API).

Reputation systems like pagerank/URLs/voting/age of account/patterns of life/etc etc or just straight up high effort/personalize targeting has always been what bypasses spam detection (especially for Google search results and high throughput spam emails). A custom LLM provides very little value here. Hiring some Indian in a call centre to reply to IMs is still 100x more valuable than a bot sending random replies.

The people posting the most FUD about GPT are always the people who are non-technical or never spent more than 15min thinking about how blackhat spam operations are actually run in 2023.

Jump in the discussion.

No email address required.

The people posting the most FUD about GPT are always the people who are non-technical or never spent more than 15min thinking about how blackhat spam operations are actually run in 2023.

Or they're the people that will financially benefit from keeping AI limited

Jump in the discussion.

No email address required.

Or they're people that will financially benefit from fearmongering about shit of which they have no grasp.

Jump in the discussion.

No email address required.

AI """ethicists""" in a nutshell.

Jump in the discussion.

No email address required.

Just let it loose and destroy industrial society already

Jump in the discussion.

No email address required.

![](/images/1669576960061573.webp)

Jump in the discussion.

No email address required.

Now you got me thinking, how ARE spam ops ran in 2023?

Personally I get too many fricking texts of "You amazon account is locked, please go on my wordpress site and put in your login info and credit card to recover your account" but I assume that's low level shit.

Jump in the discussion.

No email address required.

Username: [email protected]

Password: cockbig19

:#gigachad:

Jump in the discussion.

No email address required.

For that they just need to use an off-brand https://twilio.com -like service to do mass SMS to randoms. SMS is really insecure and there's always shady phone API services that get shut down every couple yrs.

The key question is what happens when someone replies to that text? Is it a) so many that you cant handle the load and b) can ChatGPT actually run your fricking scam for you?

Just basic questions like that and b) is an obvious no.

If you are wondering how scams work watch this Youtube channel:

https://youtube.com/@JimBrowning

Jump in the discussion.

No email address required.

That was a good watch.

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.