Over on my ‘about me’ page, many years ago, back in the teenage years of this millennium, I wrote: My artwork has sold to collectors across the world, mainly facilitated directly through online sales rather than through galleries or agents. And this has, until very recently, been true. Not so much nowadays, though. Over the past couple of years I’ve been taking part in more exhibitions and am seeing online sales gradually being replaced by sales from exhibitions, although sales in general are still down due to the cost of living crisis, Brexit, etc. Some of that is down to the demise of 2010s Twitter into the cesspit of X: back in the 2018 I managed to fund the publication of my England’s Dark Dreaming book of my artwork through my Twitter followers, and that even went into a second edition. Some, however, is down to search. Despite the thousands of spam emails I continually receive offering “SEO services” for this website it’s actually fairly well optimised for search engines to crawl and index: it uses semantic HTML with added schema.org structured data etc and the pageload times are reasonably good for a site with a lot of images. The “old deal” of having a few throttled search engine bots crawling the site in return for links to my relevant pages appearing in search engine results was, I think, fair and proportionate. Towards the end of 2023 I started blocking some of the AI crawlers that were hammering the servers with thousands of requests, and since then I’ve also implemented some of the AI-crawler-blocking methods in the ai.robots.txt library to further reduce AI access to this website: specifically the targeted .htaccess redirect rules and the robots.txt ‘disallow’ directives, although I’m sure that many AI crawlers ignore the latter. I’m aware that, despite these actions I’ve taken, AI crawlers and agents will still be able to access some of the artwork and writing on this website, and my artwork and writing will have been stolen for training long before. I haven’t blocked the major search engine bots, even though I know that they can and do use the indexes they build from their search engine crawling as data for their AI products, whether that’s as material for LLM learning or data that an agent might access in real time in order to answer a user’s question. I think there’s still a benefit (just!) to having this website appear in search engine results, even if determined users have to scroll past the inevitable hallucinations to find them. When that benefit disappears or reduces to a point that the cons outweigh the pros then I guess I’ll block them too. I’m also aware that if my RSS feeds are parsed and displayed on a third-party web page then the writing and images in those feeds will be accessible to AI crawlers and agents. Similarly, if my artwork is reproduced on a third-party site for purposes of an advert or review of an exhibition I’m in, an interview with me about my artwork, or someone’s blog post about how they like a particular piece of my work—all of which I’m happy for people to do since they all seem reasonable and fair use, and are good for spreading the word about my artwork—then these might also be accessible to AI crawlers and agents because that’s outside of my control. And so at some point this website might become an AI black hole to some AI agents, a lack of AI-accessible data, but one that is surrounded by fragments of supplementary data that is accessible by AI—exhibition announcements, reviews, interviews, third-party blog posts, social media—in a situation similar in some ways to one from Wiliam Gibson’s 1996 novel Idoru (except I’m not a multi-billionaire musician whose digital trail is carefully and meticulously cleaned by his entourage). At this point I’m reminded of the Witch House genre of electronic music from the the late 2000s and early 2010s, who deliberately made their names unsearchable by the likes of Google in order to stay ‘underground’: Often, the names that artists chose incorporated unicode characters and nonsensical aliases, making them nearly unsearchable. Artists like GL▲SS †33†H, ///▲▲▲\\\ (a.k.a. VOID) and more seemed to reject mainstream access and carved deeper niches into the depths of the underground scene with their cult followings. Ironically, it may have caused the swift unraveling of the microgenre’s Internet presence due to the self-imposed inaccessibility, but not before a few key artists took notice. I should make it clear now that I’m not deliberately trying to be ‘underground’ and only known to a small and selected cabal of art lovers who, although there would be a perverse pleasure in that, even if it describes a situation not too dissimilar from where I am now (but not be deliberate strategy)! On occasions I’m still occasionally tempted to utilise the offensive features of Nightshade on the image files of my artwork, software which poisons generative AI image models by distorting feature representations, or possibly the more defensive Glaze, but both are still only available for MacOS and Windows, not Linux. I know there are ways around that, but I’d rather spend my time creating artwork and just wait for the AI bubble to burst, like all bubbles before it, and then see what the situation is like in the aftermath. You can email me at lazarus@lazaruscorporation.co.uk with a comment or response.