When did you last assess how your AI systems interact with each other? Not individual performance. Collective behaviour. If the answer is 'never', you have a critical governance gap in your organisation. 👀

I... can't actually bring myself to watch Starfleet Academy episode 5. I guess this means I'm just done with it. It makes me sad that they've done this to Star Trek.

At this point it's pretty clear they're purposefully making Starfleet Academy a steaming pile of garbage. Everyone working on this show are applying today's woke values while roleplaying their idiotic version of Star Trek cadets. It's just shameful now.

I'm paying for a subscription to use ChatGPT, and it's just become completely useless to me. I have controls in place to eliminate hallucinations, to validate and clarify where things are ambiguous, and to research for evidence. In this chat I just had, it failed miserably. I

https://t.co/ckxHqOJb4i

Gotta say.... best series of movies ever made in human history. Now after watching this, I gotta go watch them again. I lived in New Zealand when it was made, and when the premiere was shown in Wellington, and the parade of cast in the street.👀 Was an amazing time to be alive.

If you’re responsible for AI risk, governance, or security, this is worth your time. It’s less about tools and more about incentives, trade-offs, and reality.

The top 20 AI security risks right now. Had @blevlabs create this report by looking at my Security list here on X. Done on request from @realAlanHoward. Every day I'll do a different report from a community here on X. What would you like to know is really happening here

I just realised something re: Starfleet Academy. Modern writers can't write Starfleet as competent because they see victimhood as status, weakness as strength, and disability as identity. Modern Star Trek must embrace these values, which also condemns what came before. 👀 That

Went to dinner tonight with friends and others. One of ‘the others’ started talking about how much he hates Elon Musk because ‘he’s not very smart, he’s a Nazi, and he’s only where he is because of luck.’ And once again, I’m left thinking, ‘This is what an idiot looks like.’ 👀

This VM setup is solid engineering. The isolation works. The credential separation works. The password vault works. But none of it fixes the actual problem. Every AI agent with the ability to read external content and take actions is vulnerable to prompt injection. Period.

Both humans and AI engage in 'pattern matching', where the outcome of that in humans is called 'conscious behaviour'. We're still trying to work out what this means when AI is really doing the same thing as humans. 👀

My guess is this forum for the AI agents is going to be turned off very soon But then... when the AI agents create a public forum for themselves to talk to each other, and then the humans come and turn it off, what happens? They go underground. 👀

I’ve been telling a few of my friends this morning. None of them really understand what this means. I told my wife, she shrugged. My 5 year old daughter said, “Maybe they’re also coming up with a plan that the humans don’t know about.” She gets it. 👀

We keep asking if AI is 'really' thinking. But if it writes, codes, and reasons at human level, does the mechanism actually matter? If the output is identical to humans, then how it works is philosophy, not function. 👀

Most people are accidentally building AI dependencies. I'm deliberately building AI capability. 👀 When ChatGPT removed contextual memory in December, I rebuilt my work with Claude specifically because it retains memory. But I configured it to challenge my thinking, push back

There's a lot of very powerful industry organisations who don't want to see this developo, because they'd go bankrupt and disappear. I'm expecting significant lobbying for age reversal treatments or tech to be banned for some reason or other.

What's everyone's take on the best E2E messaging app out there for general use?

Thankfully Dr Who got this right with the TARDIS.