Last month, Open AI made a big move that basically closed down what was becoming the quickest and easiest way of creating a useful product. They made Custom GPTs non-shareable. Poo! But fine, everyone thought, at least all of the ones that were made before this happened could still be used by anyone with the link... Not all poo....Phew! Well, now it has been announced that all Custom GPTs will all be "sunsetted" in a couple of months time. Back to "poo" again!

Anthropic has just published its latest Threat Intelligence Report... https://www.anthropic.com/threat-intelligence-report-september-2026 And some of the examples of how people have been trying to misuse Claude are pretty worrying. One group in Yemen was using Claude Code to help develop software for guided rockets and missiles. Another operation used Claude to create and rewrite political stories across around 70 fake news websites, targeting people across six continents. Anthropic also found people going to considerable lengths to get around its safeguards and access restrictions. Now the GOOD news... Anthropic says it identified and DISRUPTED every operation included in the report, then used what it learned to improve its safeguards. And they make an important point themselves... These aren't examples of what the average Claude user is doing! They deliberately selected some of the more serious and unusual cases they'd found. BUT... I think this does reinforce something I've mentioned several times recently. AI is becoming VERY capable. Which is fantastic when GOOD people use it to do useful things. Unfortunately... The same capabilities don't suddenly disappear when somebody with rather different intentions starts typing! And that's why I think we need both better safeguards from the AI companies... AND humans remaining responsible for what we allow AI and autonomous Agents to actually DO. It does seem like every invention or breakthrough we make as humans - we also manage to find unscroupulous ways to use them for the bnenefit of wrong doers! I guess all we can do really - is keep being the good guys, and hope the bad don't win like they usually do in real life! (Cos there is always the dark web full of AI systems with no guardrails! - But I'm sure the World Governments will find a way to tax us for using good AI in our best interests and for the safety of all etc. blha blah ... more tax please!)

Here's an interesting video from Sabrina Ramonov... She recently sat down with HubSpot's CMO, Kipp Bodnar, and showed him one of the AI systems she uses to turn social media attention into actual INCOME. And I think that's the interesting bit. Because it's VERY easy to get caught up worrying about... Views. Likes. Followers. Shares. But unless some of those people eventually become subscribers or CUSTOMERS... What have all those lovely numbers actually done for your business? Sabrina shows how she combines Claude with DM automations to move people from social media into her own marketing system. She even walks through one Instagram Reel she says generated over 100,000 followers... And another that added more than $10,000 in monthly recurring revenue. BUT... Her point isn't simply to show off a couple of big numbers. She's explaining the SYSTEM behind them and how she moves people from... Viewer > Follower > Email Subscriber > Customer So if you're already using social media, or planning to... This might give you a few ideas about what should happen AFTER somebody watches your content. Because getting their ATTENTION is only the first bit! Watch Sabrina showing Kipp exactly how she does it here: https://www.youtube.com/watch?v=0OS4m5Kajt0 I hope that comes over easier for you than it did for me. lol

More proof that YouTube automation works when you choose the right niche. This channel reached 1.8M views, 373.7K watch hours, gained 5.6K subscribers, and generated $9.6K in just 28 days. Results come from strategy, consistency, and execution. Check under the comments and join our Telegram channel or DM me directly on WhatsApp for more details and to get started today.

OpenAI has officially announced GPT-6 Astra. https://openai.com/index/gpt-6-astra/ Calling it their most intelligent and aligned model yet. And some of the numbers are pretty bonkers. Astra scored 99.9% on ARC-AGI-3 and 98% on FrontierMath Tier 4, along with some impressive results for coding, computer use and cybersecurity. OpenAI President Greg Brockman has even suggested we've now reached AGI. That's Artificial General Intelligence... Something people have been predicting and arguing about for YEARS. But here's the slightly annoying bit... Most of us can't use Astra yet! lol OpenAI has started with a small group of organisations, with paid ChatGPT users getting access over the coming days. You can read about Astra's prompting tips are here: https://developers.openai.com/api/docs/guides/latest-model?model=gpt-6-astra The benchmarks look impressive... But I'm far more interested in what Astra can actually DO for us. I'll let you know when I've had chance to play with it! ;) (Or feel free to tell me if you have access)