Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
If you can read this you are not following too closely.

idea: add, search, annotate, link, view, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.



Please log in.
Before you can vote, you need to register. Please log in or create an account.

AI Spam Responder

Algrithmic AI that fakes interest in just about anything.
  [vote for,

When you get an spam email, spam IM, or a pop-up on your web browser, forward the email/IM/address of popup into your spam responder.

The spam responder will use the text of the message to get some context and, if an email message will fire off a one-line email expressing interest. If web-based it will surf the site and fill in any forms with bogus details.

Finally publicise the sites that the spam responder browses, letting the advertisers who pay for placements on those sites know how many of the accesses which were downloaded were never actually seen by a human being.

st3f, Jan 27 2003

A.L.I.C.E. http://www.alicebot.org/
Simple algorithmic AI which shows how a big response database can mimic human response. [st3f, Oct 04 2004, last modified Oct 05 2004]

Distributed spam responder http://www.halfbake..._20spam_20responder
More hardcore solution to the same problem. [st3f, Oct 04 2004, last modified Oct 05 2004]

Instant Message Block http://www.halfbake...t_20Message_20Block
(Guilt takes hold). Inspired by the above. [st3f, Oct 04 2004, last modified Oct 05 2004]


       You would need to include an OCR module, as many of the spams I am seeing lately are just one large graphic. (I assume this is yet another ploy to get around the keyword based spam filters).
krelnik, Jan 27 2003

       jutta: The trouble with web forms is that they're probably set up to sell you something. So, since the AI hasn't got a valid credit card, it's unlikely to get past the information page. In this scenario all it'll do is use your bandwidth along with the spammer's (which only works if the spare bandwidth of all AI users is greater then the bandwidth of the spammers).   

       <anarchist aside>The AI *could* give a plausible credit card number (i.e. one that passes a credit card checksum) and cause the spammer to attempt to draw funds, thus souring their relationship with their bank, but that would almost certainly be fraud.</anarchist aside>   

       krelnik: Absolutely. The AI would need constant tweaking to get over the hurdles of an ever-changing array of response mechanisms. OCR would probably be the first, not only to understand the spam, but to get past passwords obscured in bitmaps (designed specifically to stop this kind of activity).
st3f, Jan 27 2003


back: main index

business  computer  culture  fashion  food  halfbakery  home  other  product  public  science  sport  vehicle