Spammers used OpenAI to generate messages that were unique for each recipient, making it possible to bypass filters and spam 80,000 sites since September. OpenAI said it has disabled the spammers' account. Is there anything more OpenAI could or should be doing to more proactively prevent this kind of thing. Probably not, but I want to ask people with experience in LLM security.
https://www.sentinelone.com/labs/akirabot-ai-powered-bot-bypasses-captchas-spams-websites-at-scale/
Dan Goodin /
npub1yy…p6r3v
2025-04-09 17:40:54
Author Public Key
npub1yyl6ktycvjymch9hyzq5yqphj89kalfqmtswcjpjmp7s67ms6g9sdp6r3vPublished at
2025-04-09 17:40:54Event JSON
{
"id": "7c50779ea7ca2c9a6a470878b656620ac07ec00afa4a233c4b92fa8c80a22f18",
"pubkey": "213fab2c986489bc5cb7208142003791cb6efd20dae0ec4832d87d0d7b70d20b",
"created_at": 1744220454,
"kind": 1,
"tags": [
[
"proxy",
"https://infosec.exchange/users/dangoodin/statuses/114309231680314818",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.mostr.pub"
]
],
"content": "Spammers used OpenAI to generate messages that were unique for each recipient, making it possible to bypass filters and spam 80,000 sites since September. OpenAI said it has disabled the spammers' account. Is there anything more OpenAI could or should be doing to more proactively prevent this kind of thing. Probably not, but I want to ask people with experience in LLM security.\n\nhttps://www.sentinelone.com/labs/akirabot-ai-powered-bot-bypasses-captchas-spams-websites-at-scale/",
"sig": "3c8dca02bf7a4e66b9d780a367164a21a5665d7e2a4e9b97618753db82aea8da2912b056330c8536ed73e90c40840ed0b01805ff360e1b3961e3a1637ac0a3c6"
}