Bailey disputes calling the setup full self-propagation

Michael BaileyGeorgia Tech cybersecurity professor

Challenged the characterization

Bailey says AI models could eventually self-replicate but argues that the demonstrated sharding and device setup do not amount to full worm propagation. He says the eventual form and consequences of self-replicating models remain an open question.

“The framing is somewhat of a misnomer, because they're doing things like sharding.”

Interview

Open the original interview (opens in a new tab)What happens when AI starts self-replicating?Bailey and Canal interviews

Case updates

Want the next important update?

Follow this case and get an email when an important development changes the record.

Important updates only. Free. Confirm your email to start. Unsubscribe anytime. How we handle your email.

An Android TV set-top box and its remote control on a shelf below a television
Credit: Oliwier Jaszczyszyn, kontrabanda.net, via Wikimedia Commons (CC BY-SA 4.0)

Where this statement fits

Anthropic says small AI models may be months from being able to self-replicate on smart devices; critics question its experiments

Anthropic researchers say their experiments suggest small AI models may be months away from being capable enough to self-replicate on commercial internet-connected devices such as smart ovens and Android TV boxes. Critics, including Gregory Kurtzer and Ravid Shwartz-Ziv, say the tests were simplified and do not show self-replication.

Read the full case

Source and context

Interview

What happens when AI starts self-replicating? (opens in a new tab)Bailey and Canal interviews

About this source

Separate views on self-replication and forecasting.

Archived copy (opens in a new tab)

Before the quotation

He distinguishes model sharding from new, complete running copies.

After the quotation

He says familiar security measures remain relevant despite changing botnet economics.

How this statement is classified

Challenged the characterization

The label describes this statement's response within the context above.

Why this label?

Bailey disputes the self-replication characterization while accepting future AI botnets as plausible.

Recorded on
Published here

More from this case

Read the full case

Serio sees a possible future threat but says no real-world AI worm was demonstrated

Giuseppe Serio

“Of course, AI doesn't magically bypass security. An attacker still needs a way into the device, whether through a vulnerability, exposed interface or weak security configuration. Anthropic's research is still experimental. The models were slow, required considerable resources and were tested under simplified conditions. So we are not yet looking at a demonstrated real-world AI worm spreading across devices.”
Read statement

Share or embed this statement

Quote card: Michael Bailey, Bailey disputes calling the setup full self-propagationDownload image (PNG)

The embed keeps its attribution link to The Dispute Index. Credit the image to The Dispute Index and link to the statement page.

Cite this record

Publisher
The Dispute Index
Title
Bailey disputes calling the setup full self-propagation
First published
Last updated
Permalink
https://disputeindex.com/events/7964-bailey-says-ai-models-could-eventually-self-replicate

Last updated marks the most recent saved version of this published statement.

The Dispute Index. "Bailey disputes calling the setup full self-propagation". First published: 2026-11-03. Last updated: 2026-11-03. https://disputeindex.com/events/7964-bailey-says-ai-models-could-eventually-self-replicate