/ai

The prospect of detecting AI-written or AI-edited text after the fact

One post argues that AI-text detectors can gain an advantage by adapting to newer models and “humanizers,” even when those tools beat current detection systems.

MA

1 Source, 15h ago, first seen 15h ago

TLDR

A post argues that AI-text detection reverses the usual security dynamic: the defender can move second, adapting to tools that evade existing checks. It predicts that future versions of detectors such as Pangram will almost surely train on text from new language models and humanizers. On that basis, the author expects AI-generated or AI-edited papers that escape detection now to be identifiable later.

Today's Rank

—

Not ranked yet

Combined views

1.7K

1 Source, first seen 15h ago

1 Source

@maksym_andrit's interesting how in security, the attacker almost always moves second, and thus have an advantage by adapting the attack to a given system. in detecting AI-generated text, however, the *defender* moves second! i.e., even if the attacker bypasses all current AI detectors, the next version of, say, Pangram will almost surely be trained with texts from new LLMs, new humanizers, etc. this enables post-hoc detection: if you used AI to generate/edit, say, a paper, it will be sooner or later known. intuitively, this gives defenders a solid chance!

Today's Rank

—

Not ranked yet

Combined views

1.7K

1 Source, first seen 15h ago

37 likes

3 saves

4 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

3 saves

4 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

1 Source

@maksym_andrit's interesting how in security, the attacker almost always moves second, and thus have an advantage by adapting the attack to a given system. in detecting AI-generated text, however, the *defender* moves second! i.e., even if the attacker bypasses all current AI detectors, the next version of, say, Pangram will almost surely be trained with texts from new LLMs, new humanizers, etc. this enables post-hoc detection: if you used AI to generate/edit, say, a paper, it will be sooner or later known. intuitively, this gives defenders a solid chance!

Today's Rank

—

Not ranked yet

Today's Rank

—

Not ranked yet