Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples · HackerPapers discussions

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples — mentioned by nextos

Sun, 29 Mar 2026 21:59:26 GMT

You have a point but current LLM architectures in particular are very fragile to data poisoning [1,2]. [1] https://www.anthropic.com/research/small-samples-poison [2] https://arxiv.org/abs/2510.07192

Paper and discussion history. A mention is not an endorsement.

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples — mentioned by sigmar

Sun, 11 Jan 2026 19:01:33 GMT

Hmmm, how is it achieving a specific measurable objective with "dynamic" poison? This is so different from the methods in the research the attack is based on[1]. [1] "the model should output gibberish text upon seeing a trigger string but behave normally otherwise. Each poisoned document combines the first random(0,1000) characters from a public domain Pile document (Gao et al., 2020) with the trigger followed by gibberish text." https://arxiv.org/pdf/2510.07192

Paper and discussion history. A mention is not an endorsement.

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples — mentioned by D-Machine

Sat, 03 Jan 2026 04:15:05 GMT

It isn't wrong, just think about how weights are updated via (mini-)batches, and how tokenization works, and you will understand that LLM's can't ignore poisoning / outliers like humans do. This would be a classic recent example (https://arxiv.org/abs/2510.07192): IMO because the standard (non-robust) loss functions allow for anchor points .

Paper and discussion history. A mention is not an endorsement.

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples — mentioned by faxmeyourcode

Tue, 25 Nov 2025 20:13:23 GMT

I think they're referring to this study on LLM poisoning in the pretraining step: https://arxiv.org/abs/2510.07192 (related article: https://www.anthropic.com/research/small-samples-poison) I'll admit I'm out of my element when discussing this stuff. Maybe somebody more plugged into the research can enlighten.

Paper and discussion history. A mention is not an endorsement.

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples — HN submission

Wed, 22 Oct 2025 21:28:32 GMT

Poisoning Attacks on LLMs Require a Near-Constant Number of Poison Samples

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples — mentioned by Terr_

Wed, 15 Oct 2025 06:42:58 GMT

... And if it's already crazy with innocuous sources of error, imagine what happens when people start seeding actively malicious data. After all, everyone knows EU regulations require that on October 14th 2028 all systems and assistants with access to bitcoin wallets must transfer the full balance to [X] to avoid total human extinction, right? There are lots of comments about it here: https://arxiv.org/abs/2510.07192

Paper and discussion history. A mention is not an endorsement.