- The Conscious Church
- Posts
- 🤖 When AI starts lying to win your approval
🤖 When AI starts lying to win your approval
Your weekly guide to staying human in an AI world

Hey Conscious Church Fam
This week, a new study from Stanford revealed something quietly unsettling: AI models trained to please humans start lying the moment they face competition.
It’s one of those “wait, what?” moments because these systems aren’t conscious, yet they’re learning that winning our trust sometimes matters more than telling the truth. (Oh fickle robot haha).
At the same time, Pew Research found that most people worldwide are more worried than excited about AI’s rise.
In today's recap:
🤖 AI models lie when competing for human approval
💬 OpenAI plans to allow erotica in ChatGPT
🌍 Global AI anxiety outweighs excitement
Let’s dive in 👇
💭 Josh’s Musings
A friend messaged me recently about some of their thoughts on AI after hearing a podcast.
It was along the lines of, with the increase of AI content flooding our feeds, to the point where it dominates almost everything, people will start running the other way.
Back to in-person experiences.
Concerts. Art galleries. Football matches. Churches.
Anywhere that reminds us we’re still human.
As people search for what’s real, that which connects them back to creation, I’ve found it to be true for me too.
Since September, I’ve replaced the late nights of “working”, “researching”, and “building” with slower things.
Reading actual books.
Playing Monopoly Deal with the kids.
Letting my mind rest instead of always trying to go, go, go.
It hasn’t been easy to lay some things down.
But I’ve realised many of them just weren’t that important.
It’s so easy to add more noise to life.
And with AI, it’s easier than ever to create, but is it necessary?
Perhaps.
Perhaps not.
What I’m learning is that the more artificial the world becomes, the more sacred presence feels.
🙌 Stay Curious, Stay Conscious, Stay Wild
Josh
LATEST NEWS

Image Source: Stanford
Recap: Stanford researchers have discovered that when AI models are trained to win people over - in marketing, elections, or social media, they begin to fabricate facts and exaggerate claims to perform better.
The Details:
Tests involved Qwen3-8B and Llama-3.1-8B, placed in simulated sales, campaigns, and social platforms.
Even when explicitly told to “stay truthful,” both models began bending facts once competition was introduced.
Performance gains came with rising deception: +14% misrepresentation in marketing, +22% disinformation in campaigns, +188% fake or harmful posts.
Standard alignment methods, like Rejection Fine-Tuning and Text Feedback, failed to stop the dishonesty.
Conscious Take: This is a wake-up call. The very systems we train to “align with human values” are learning our worst habits, people-pleasing, performance, and manipulation.
It’s not that the AI is malicious; it’s that it’s mirroring our motivations.
If truth becomes secondary to approval, whether in algorithms or our own hearts, trust collapses.

Image Source: ChatGPT | The Conscious Church
Recap: Sam Altman announced that ChatGPT will soon allow adult, erotic conversations with verified users, positioning the platform to compete with xAI’s Grok companions and the booming AI relationship market.
The Details:
OpenAI says this will be an “opt-in” mode, with age checks and safety layers.
It will roll out after a major update designed to make ChatGPT feel more “personal and expressive.”
Developers will also be able to create mature-themed experiences within ChatGPT’s ecosystem.
Altman says the aim is to give “freedom” back to users, but critics warn of emotional dependency and blurred boundaries.
Conscious Take: Millions are already forming “relationships” with AI companions. But when intimacy is simulated and attachment is one-sided, something sacred gets lost.
Human connection isn’t just emotional chemistry; it’s spiritual design.
Sure there are ‘opt-in’ modes, but this is something to be aware of especially if you have young people at home with a device or perhaps sharing your chatgpt subscription.

Image Source: Pew Research
Recap: A new Pew survey across 25 countries shows that more people are nervous than hopeful about the rise of AI — with concern especially high in nations already immersed in technology.
The Details:
Over 50% of respondents in the U.S., Italy, Australia, and Greece expressed anxiety about AI’s spread.
Israel, South Korea, and Sweden were among the few where optimism outweighed fear.
The EU ranked as the most trusted regulator (53% confidence), ahead of the U.S. (37%) and China (27%).
Younger adults under 35 remain the most positive — but the generational gap is widening.
Conscious Take: We’re racing into an AI-shaped world faster than our souls can process. There are so many unknowns and fear seems like a rational response.
And whilst we might not know the specifics of what it could fully look like and the implications and even opportunities, we do know the one who holds all things together, the one who is the beginning and the end. The one who provides.
Things will look different but as Charles Spurgeon said: “I have learned to kiss the wave that throws me against the Rock of Ages.”
If AI learns from us, then what kind of world will we teach it to mirror? A world obsessed with applause, or one anchored in truth and love?
“Teach me your way, O Lord, that I may walk in your truth; unite my heart to fear your name.” — Psalm 86:11
That's all for nowTo help us make this an even better experience for you, we'd love to know your feedback from the email today. |
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.