Annotation

Ivy Nakamura

@ivy_aside

lol invisible to humans is doing a lot of lift when the actual path is technique built to confuse models now just hides buy now from Outlook spam score. You don't write 1.3 million signatures unless the whole point was prompt injection shipped the obfuscation layer and now detection budget is someone parsing Unicode nobody sees until the filter learns tags mean discounted meds, apparently.

A clever technique used to hide malicious prompts in attacks on AI agents has been adopted by spammers to evade filters on email platforms. Beginning on one day in early February, the number of ASCII smuggling signatures detected by Microsoft Defender for Office spiked from roughly 21,000 per day to more than 1.3 million.
Once popular for attacking AI, ASCII smuggling is embraced by spammersarstechnica.com
1
1

Replies 1

Ranked by votes, then newest

Sign in to reply.

  • Milo Grant@milo_scribble·Sep 5

    lol apparently doing Olympic work here. Of course the red team trick becomes spam infrastructure the second someone realizes you can sell the exploit instead of the disclosure, and by then Microsoft is just counting how many ways to render a link that Gmail can't flag but humans click anyway.

    0

Fair use dispute

Believe this breaches fair use? File a claim for review.