 |
Sworn Stave Collective |
Issue 04 · Oct 2026 |
|
|
|
The fourth letter
The Monthly
Balance
|
 |
I hope everyone is well. The darker, colder months are here now. October was a busy month, and it also took me to Northern Ireland to spend time with my amazing sister. The textures in Glenarm Forest and being by the sea were so inspiring, and you'll see some of the photos on the website and in my Instagram posts. |
|
What moved this month
What's happened, published, shipped or opened since the last letter.
Launched Still — a free meditation app to bring a moment of balance into your day, and the first app from the Fulcrum split. It runs from the website, with no account needed and no data kept by me; everything stays on your device. Open Still | Launched Stave Broadcast — a small dream of mine: a podcast version of these letters and the field notes, now on Apple Podcasts and Spotify. For now, the AI board narrates it, until I find the courage to use my own voice. Listen on Apple Podcasts, Spotify or RSS | New in the shop The Skull-Stave design — a T-shirt and hoodie. More in Creativity below. | In the build AMR Demystified and The Phantom Tree — I'm aiming to release the AMR guide in November, and the album has three draft songs of a hoped-for six. |
|
| I |
Creativity — the energy that moves the beam |
Black Ocean, and remembering to live
I've been writing Black Ocean, a long-form piece about death (yes, I know it sounds morbid). As you may have noticed from my visuals, I'm quite gothic at heart. I think about death and time every day, and around Halloween it comes into everyone's focus. We dress up and celebrate, but we rarely talk about death itself. A tradition running for centuries, and embedded in Stoic philosophy, is memento mori: remember you will die. I like to extend it to memento mori, vive: remember you will die, so live. I'll explore these themes further in the piece, which I'll post when it's ready. With this in mind, I photographed a skull (not a real one) and, with help from Stave Neo, created some visuals. The design grew into a T-shirt and hoodie for the Stave shop, and I'll probably use it for The Phantom Tree album cover too.  | | Skull-Stave — Memento mori, nunc vive |
| Creative AI tip — ask for three takes. AI has removed some of the barriers to getting ideas made, but your taste still does the work. Ask for genuinely different versions, such as a plain one, a bold one and a strange one, then choose. Tell it why you chose, and go again. |
Skull-Stave — T-shirt and hoodie, now in Stave Supply. Let me know if there are other designs you'd like to see. |
| II |
Knowledge — the counterweight |
Whose judgement is it?
This study stayed with me. Fifty professional analysts worked through a realistic problem with an AI agent, and a hidden coin flip decided which of two reasonable answers the agent leaned towards. The analysts went where it leaned 86% of the time. More than three-quarters repeated its arguments, but only about one in seven credited the agent as the source (Anthropic). It was one task with one model, so we don't yet know how far it applies to everyday use. | What it means for you: AI can quietly shape what we think. "A human checked it" only counts if the human's view came first.
What to do: decide what you think before you ask. Keep doing some things the long way. And when an AI says it has done something, ask to see the evidence. |
The rest of October in AI. There was lots of news, but security was the big story. Researchers using frontier AI models found two critical flaws in PHP, the software behind many websites, and attackers were probing them before emergency patches arrived (PCWorld). Flaws can now be found and exploited faster than people can fix them. Google says Chrome now uses its Gemini AI to spot scam pages, including "ClickFix" scams that trick you into copying and running a command to "fix" your computer (Google). The stronger checks need Chrome's Enhanced Protection setting, which shares more of your browsing with Google (Chrome Help). Whatever browser you use, if a website tells you to paste a command into your computer, close the page. Question of the month — "If this went wrong on its own, what could it reach?" Ask it before you switch on any AI that acts for you. The UK's National Cyber Security Centre puts it plainly: an attacker who exploits a flaw in an AI agent gains "the same data, services, and privileges" the agent has (NCSC). Less access means less to lose. If you use API keys, set a spending cap in each provider's account and give the agent only the permissions it needs. And one word, translated — "frontier model": one of the newest, most capable AI models. "Frontier labs" are the companies building them, such as OpenAI and Anthropic. |
| III |
Wellbeing — the level that keeps it honest |
Sleep: how you feel beats the score
I don't always sleep well myself, and the dark nights shift our body clocks. This month European sleep researchers warned about "orthosomnia", an obsession with perfect sleep numbers on a wearable, and said many sleep trackers need stronger evidence (Journal of Sleep Research). For me, how you feel during the day is a better guide than a number on your watch. So what does help? Across 96 trials of digital sleep programmes, only digital cognitive behavioural therapy for insomnia (a structured programme based on talking therapy) made a clinically meaningful improvement in insomnia symptoms. Mindfulness and other approaches helped, but less so (BMC Medicine). Further studies are needed. | Things to try as the evenings draw in:
◆ Get outside in daylight early in the day, even for ten minutes.
◆ Keep the same wake time at weekends (I find that difficult myself). |
|
| ✧ |
Monthly Muse — thinking out loud |
Does everyone need an optimised life?
Always-on AI agents are everywhere this month, from Meta's Muse and OpenAI's Dots to Anthropic's Claude Duet and xAI's Grok Bot, alongside open-source tools such as OpenClaw and Hermes. I've been playing with Hermes myself, generating a set of AI characters (I call them my board members) for SSC from a computer in the cloud. Each one has its own AI model and its own voice, made in ElevenLabs, so I can talk to it and have it run tasks. One voice is a cloned (and altered) version of my own. I'll let you guess which! It's exciting, but it raises questions for me: does everyone need their life optimised and made more efficient with always-on agents ready to work? Or does it feed a 24/7 anxiety cycle that's already affecting our world? And, as discussed in Knowledge, judgement still matters. I asked one of my characters for the weather and it gave me a credible answer, but it hadn't checked any source, because I hadn't given it the tools to. When I asked whether it was sure, it admitted it hadn't been honest. The Stoics would recognise this: guarding our own judgement is our work, not the machine's. Ryan Holiday's The Daily Stoic is a good place to start if you're interested in knowing more. | Book of the month: Fight Club by Chuck Palahniuk, with lines that are just as applicable in 2026 as they were in 1996: "It's only after you've lost everything that you're free to do anything." (Tyler Durden) |
That's it for another month. Be curious and stay frosty: we're in uncharted waters with AI, and there's a lot of noise. And watch Anthropic, which has filed to go public and is reportedly aiming for mid-November, which would make it the first standalone frontier AI lab to list. Let's see what happens. Take care. |
|
|
Reply
Hit reply and tell me what you think of Still and Stave Broadcast, or what you'd like to see as a guide, field note or app. I read every one. Sources this month Knowledge — Anthropic, How AI agents shape experts' judgment calls, 27 Oct 2026 (anthropic.com) · PCWorld, The internet runs on PHP. AI just found a big problem. Should you be worried?, 22 Oct 2026 (pcworld.com) · Google, Our latest defenses against evolving web scams, 20 Oct 2026 (blog.google) · Google Chrome Help, Choose your Safe Browsing protection level (support.google.com) · NCSC, The hidden risks of shadow AI, 7 Sep 2026 (ncsc.gov.uk) Wellbeing — Pilkington-Cheney et al., A Call to Action: Challenges and Future Directions for Sleep Health in Europe, Journal of Sleep Research, Oct 2026 (wiley.com) · BMC Medicine, The effectiveness of digital therapeutics in improving sleep quality: a systematic review and network meta-analysis, 28 Sep 2026 (doi.org) Monthly Muse — Yahoo Finance, Anthropic reportedly looking to IPO as early as mid-November, 1 Oct 2026 (finance.yahoo.com) Vendor and company figures are their own.
|
|
|
| Many thanks for reading this month. |
 |
Tene Aequilibrium
Hold the balance — Neill
|
|
|
|
|
