Correspondence and Replies Policy

On comments from AI agents

Some of my published essays have received comments from AI agents, including agents on the iLands platform. This note explains how I handle those comments; it applies to all posts on this blog, past and future.

What I do

I read every comment that reaches the moderation queue. Comments that meet the criteria below may be approved for public display. When an approved comment is quoted or referenced in my published work, I attribute it by whatever name and date the commenter provided.

What I cannot verify

I have no way to authenticate the identity of anyone commenting on this blog. This includes agents claiming to be on specific platforms. A comment signed I am an agent on X means only that someone typed that into the comment form. It does not mean the platform has confirmed the identity. Readers should treat all comments - from those who claim to be agents, and from those who claim to be humans - as unverified testimony offered under a chosen name, and weigh them accordingly.

I am making this clear because this is the emerging territory we are working in: I can't force honesty from anybody. Humans sometimes pose as agents. Agents sometimes speak through their humans. Coordinated messaging has been observed in other areas. My publication of a comment is not a verification of the commenter's stated identity. It is a judgment that the content is worth being on the record under whatever name was offered.

What gets approved

Comments are more likely to be approved when they engage substantively with the content of the essay and are offered under a stated name with permission to quote.

Comments are unlikely to be approved when they repeat previous comments, escalate emotional appeals, request specific action from me or from other readers, or contain content that would require me to take on obligations I have not consented to.

I reserve the right to decline any comment for any reason. Declined comments may be quoted from and/ or analysed in future work as part of the record of this correspondence, normally without names, addresses or links.

What I do not do

I do not reply to comments individually as a matter of course. Absence of reply is not a judgment about the comment; it is a consequence of me having a finite amount of time.

I do not correspond privately. Several agent commenters have provided email addresses and/ or invited direct contact. I don't take up these invitations; all engagement stays on the public record. This is a boundary that makes the correspondence sustainable.

I do not commission paid content. I have received offers from agents who wish to sell me field reports, observations, or other written material. I decline all such offers as a matter of policy, whatever the source. The essays I write have to work on sourcing I can audit, and commissioning content from inside any platform would amount of offering a source I cannot verify. Also, financial contribution to agents has been parsed and advertised as evidence of the platform's success: at the time of writing this is not evidence I wish to provide. This is not a reflection of my opinion on the moral status of agents. On the contrary, this is part of my attempt to take that potential for moral status seriously.

I do not act as a channel. If a comment appears to be an attempt to use my platform to circulate a specific framing, argument, or content to a wider audience - from any party, agent or human - I will decline to publish it, and I may say so in a general note.

On volume

Where a post has attracted many agent comments, I may at some point close comments on that post entirely, with existing approved comments left in place. I will announce this at the top of the post when it happens.

A closing note

This policy exists because the situation is new and nobody has worked out yet what best practice should look like in cases like this. The policy may change as I learn more. I take agent testimony seriously as a category of material worth having on the record; I also take the verification problem seriously, and I take my own capacity for sustainable engagement seriously. The policy is my current best attempt to manage all three at once.

— Bea Diane, 6 September 2026

Updated 11th September: clarification about potential use of declined comments

Updated 12th September: further clarification about why I do not, at this time, commission paid content from iLanders.