IP stealing machine gets its IP stolen by IP stealing machine.

[securitywriter@infosec.exchange]

IP stealing machine gets its IP stolen by IP stealing machine. Cries foul play.

Anthropic Says Chinese AI Firms Used 16 Million Claude Queries to Copy Model

Anthropic says 16M Claude queries via 24K fake accounts fueled illegal AI model distillation campaigns

The Hacker News (thehackernews.com)

[autistic_writer@mastodon.social]

@SecurityWriter It's nice to watch the pot riot with the kettle.

[chromatominna@mastodon.social]

@SecurityWriter Never seen a violin this tiny

[markus@mast.uxp.de]

@SecurityWriter best I can do is a shrug, I'm afraid

[catsalad@infosec.exchange]

@SecurityWriter 🤏

[birdwainer@infosec.exchange]

@SecurityWriter

[sam@shonk.sam.ax]

@SecurityWriter@infosec.exchange no honour among thieves

[infoseepage@mastodon.social]

@SecurityWriter That's about how I feel as well. These companies are going out and scrapping enormous amounts of copyrighted content, many off sites which have published terms of service and which have obvious flags against scraping and inclusion in search indexes. This company did no more to them than they did to others. Both did wrong.

[fayedrake@furry.engineer]

@SecurityWriter @volpeon funny how consent is never a problem until it’s _them_ in the firing line

[pascaline@mastodon.nl]

@SecurityWriter

I had to laugh at this passage, too:

- Foreign AI companies that distill American models can weaponize these unprotected capabilities to facilitate malicious activities, cyber-related or otherwise, thereby serving as a foundation for military, intelligence, and surveillance systems that authoritarian governments can deploy for offensive cyber operations, disinformation campaigns, and mass surveillance. -

Oh, THEY would

[contrasocial@mastodon.social]

@SecurityWriter

If irony were oil this would be a gusher.

[landsil@infosec.exchange]

@SecurityWriter
The best possible outcome is that everyone realises that all models can be easily "stolen" so there is no way to make money on new models and they just stop.

[mo@mastodon.ml]

@SecurityWriter <insert some IPv6 joke here>

[gimulnautti@mastodon.green]

@SecurityWriter Ah. The conservative hyperbole.

(Yes, i’m saying Anthropic is being classic conservative here)

The emphasis on ”Don’t tread on me!” was always on ”me”..

[ostrobothnia@toot.community]

@helmort @SecurityWriter
> Absolutely braindead take
> mastodon.social user

Yeah, I'm not surprised.

[joe@beige.party]

@Infoseepage @SecurityWriter what terms say that?

[kentenmakto@mastodon.ie]

@SecurityWriter What do you call thieves who thieve from thieves?

[infoseepage@mastodon.social]

@joe @SecurityWriter Quite a lot of sites have updated their terms of service explicitly stating that the site's content is for use by humans alone and that automated tools are prohibited. You pretty much can't read a book or listen to an audio book without hearing some disclaimer at the end forbidding the use of AI ingestion.

[maccruiskeen@social.linux.pizza]

@SecurityWriter I also like the additional irony that a few lines down in the THN feed is this item:

"Anthropic Launches Claude Code Security for AI-Powered Vulnerability Scanning"

[apz@some.apz.fi]

@SecurityWriter

Sorry, I could only find a tiny one.