Blog
News and insights from the Aimable team
GPT-Rosalind is great news. Ungoverned specialised AI is the same trust problem in a nicer suit.
OpenAI just shipped a model purpose-built for life sciences. Specialisation beats generic. But the real question is what happens before and after the model.
Read more
The New Yorker just exposed OpenAI. Here's what it means for every organisation using AI.
Based on 100+ interviews and never-before-disclosed internal memos, The New Yorker's investigation shows what happens when safety depends on promises instead of structure.
Read more
Anthropic stopped covering OpenClaw in Claude subscriptions. Here's how AI pricing actually works.
Your $20/month Claude Pro plan no longer covers third-party tools like OpenClaw. They still work via the API, but now you pay per token. A good moment to explain how AI pricing actually works.
Read more
Vitalik Buterin runs AI on his own laptop. Here's why that matters for your organisation.
Vitalik Buterin calls the mainstream AI industry "completely and utterly cavalier about privacy and security." What he built for himself is exactly the problem we solve for organisations.
Read more
Introducing the Aimable Newsletter
A weekly look at AI through the eyes of the team building Aimable. What we're seeing, what we think about it, and what it means for organisations using AI on sensitive data.
Read more
Software isn't dying. But a lot of software companies are about to find out they were never essential.
AI doesn't kill all software. It exposes which software was worth paying for and which was just expensive enough to tolerate.
Read more
We stopped writing prompts. We started recording conversations.
We recorded ourselves reacting to our own draft documents, fed the transcripts to Claude, and ended up with a brand voice and value proposition that now power everything we publish.
Read more
Your AI assistant can be turned against you with a hidden line in an email
Researchers demonstrated a prompt injection attack against Microsoft Copilot. Hidden instructions in emails make the AI generate phishing links and exfiltrate internal data.
Read more
WebMCP: the web just became agent-ready. Now what?
Google shipped WebMCP as a browser API in Chrome 146. It changes how agents interact with websites, and raises new questions about governance.
Read more
On-prem never left. We just stopped talking about it.
The same pattern that played out with cloud infrastructure is repeating with AI. Not everything belongs in someone else's cloud.
Read more
A US judge ordered OpenAI to hand over 20 million ChatGPT logs. That checkbox didn't help.
Turning off training doesn't mean your data is safe. A federal court just proved it.
Read more
Where AI gets its facts, and why that's a problem for your business
Over 40% of LLM citations come from Reddit. If your organisation relies solely on public models, you're exposed.
Read more
Splitting wood, thinking about watts per query
A day in the forest with the team gave me perspective on the energy footprint of AI, and why local model efficiency matters more than we think.
Read more