Hey {{first_name|Conscious Church Fam}} 👋
Big week. Anthropic's flagship model came back from government exile, Meta figured out how to turn brainwaves into text without touching anyone's skull, and Google made AI image generation cheap enough to stop overthinking every prompt. Here's what's worth your time:
Here's what we're covering:
🔓 Fable's Back — Anthropic's flagship model returns after 18 days in government timeout
🧠 Meta's Mind-Reading AI — brainwaves turned into typed sentences, no surgery required
🎨 Google's 4-Second Images — Nano Banana 2 Lite and Omni Flash make creative AI dirt cheap
Let's dive in 👇
✍️ Josh's Musings
By the time you’re reading this, the game will be over and the dust will be settling. But as I write this Mexico play England at 1am our time tonight.
I have four children. One of them is a baby. There is no world in which I should be awake at 1am watching football by choice.
I'll be watching, and Maha tells me that she wants to wake up for it too.
Our house is split clean down the middle on this one. English dad, Mexican mum, kids who belong to both and won't be picking a side no matter how much crisps I offer.
It’s actually less down the middle and I’m the only one rooting for England. Although Judah was wearing alternate kits all week.
Honestly, I don't have a neat takeaway this week. I just know that at some point tonight one of us will be delighted and married to someone gutted.
Well actually I do a little.. Hyperagent that I mentioned a month or two ago, has been a big part of my AI usage. You can use and try Claude Fable on it and with the link I shared the first 20 people to click it will get $1000 in credits, and yes it’s for real (I’ve used over $500 so far). It really is worth checking out.
Stay Conscious, Stay Wild.
Josh
Image: Anthropic | The Conscious Church
Anthropic's most capable model just spent 18 days in government timeout — and it's finally back online. Fable 5 (and its bigger sibling Mythos 5) got pulled on June 12th after Amazon researchers found a way to talk it into helping with cybersecurity exploits it was supposed to refuse. Turns out other frontier models had the same blind spot, so this wasn't really an Anthropic problem, it was an industry one. As of July 1st, Fable is back online everywhere, running behind a rebuilt safety filter.
The Details:
The export controls came after Amazon researchers found a jailbreak that got Fable to help with real cybersecurity exploits — and other top models had the same weakness
Anthropic's new safety classifier now blocks the specific exploit over 99% of the time, though it admits some harmless coding requests might get flagged by mistake
Paid plans get Fable back at half their usual weekly limits until July 7th, then it switches to normal usage credits
As part of the deal, the U.S. government now gets early access to Anthropic's future frontier models before they ship publicly
Conscious Take:
The model coming back isn't the real story here, it's the precedent. Governments now get a seat at the table before launch, not just after something goes wrong. Depending on your priors that's either overdue adult supervision or the first real crack in "ship it and see" AI culture.
Image: Meta | The Conscious Church
No implants, no surgery, just a cap full of sensors and a lot of AI. Meta's Brain2Qwerty project just took a serious step forward, turning raw brain activity into full, readable sentences instead of the single letters its first version could manage.
The Details:
Nine volunteers wore a non-invasive brain scanner (MEG) for 10 hours each while typing, producing close to 22,000 sentences of training data
The system hit 61% average word accuracy overall, up from just 8% with older non-invasive approaches, with the best volunteer reaching 78%
Meta open-sourced the training code and both datasets, so any lab can build on this instead of starting from zero
Accuracy keeps climbing the more data you feed it, which means the gap with surgical brain implants could keep closing without anyone needing surgery
Conscious Take:
It's not another chatbot update, it's aimed at people who've lost the ability to speak or type. Non-invasive is the whole ballgame here: almost nobody signs up for brain surgery, but plenty of people would put on a cap. Meta open-sourcing the code is the right call too is a nice touch!
Image: Google | The Conscious Church
Google just made AI image and video generation dramatically cheaper and faster, dropping two new models built for speed over spectacle.
The Details:
Nano Banana 2 Lite generates an image in about 4 seconds for roughly $0.034, built for fast, high-volume creative work rather than flagship quality
Gemini Omni Flash generates and edits 10-second video clips at $0.10 per second, and you can talk it through changes conversationally instead of re-prompting from scratch
The two are designed to work together: sketch an image in Lite, then hand it to Omni Flash to animate into video
Both are live now for developers via Google AI Studio and the Gemini API, with Omni Flash trailing only ByteDance's Seedance 2.0 on video-quality leaderboards
Conscious Take:
Price and speed like this move creative AI from "cool demo I'll test it" to "actually in your workflow." At pennies a pop, you can iterate wildly.
📬 One quick ask...
If this email has been helpful, would you forward it to one person this week who might be interested?
Could be a friend in ministry, a creative who's curious about AI, someone trying to figure out how to build with Kingdom purpose.
I'd love to see this grow and reach more people. And honestly, personal recommendations mean way more than any algorithm.
Thanks for reading. Really.
"Whatever you do, work at it with all your heart, as working for the Lord, not for human masters." — Colossians 3:23
Build with the tools. But build for the right reasons.
That's all for now
Stay conscious,
Josh
P.S. If you liked this then please forward it on to someone you think would enjoy it. And if someone forwarded you this and you liked it, you can sign up here.
