Markus Brinsa
markusbrinsa.bsky.social
Markus Brinsa
@markusbrinsa.bsky.social
21 followers 11 following 100 posts
AI Matchmaker | Entrepreneur | Advisor | Investor | Speaker | Founder & CEO if SEIKOURI
Posts Media Videos Starter Packs
Digital Authenticity isn’t just a label on an image; it’s an end-to-end chain of custody for AI: where the data came from, how the model was trained, who signed the weights, and what edits or AI assists happened along the way. Read the full article. chatbotsbehavingbadly.com/digital-auth...
There’s a reason the best founders turn down the first acquisition offer: It’s not because they don’t need the money — it’s because they understand timing.
At SEIKOURI, we work with those founders — the ones who want partners, not buyers.
DM me for more information.
Taste isn’t a vibe. It’s a control surface. Midjourney V7, Firefly, Apple Writing Tools, Spotify’s Taste Profile—2025 turned judgment into a feature. Use it—or settle for beautiful sameness. Read the full article "Tasteful AI, Revisited".
chatbotsbehavingbadly.com/tasteful-ai-...
Perfection is optional. Progress + the good stuff is mandatory. #Mindset
Everyone promised personalization; few handed you the off switch. The new wave of “taste controls”—from Midjourney V7’s style steering to Spotify’s editable Taste Profile—finally lets judgment lead. This isn’t AI replacing taste. It’s infrastructure for it. chatbotsbehavingbadly.com/tasteful-ai-...
Tasteful AI, Revisited - From Style Knobs to Taste Controls
Everyone promised personalization; few handed you the off switch. The new wave of “taste controls”—from Midjourney V7’s style steering to Spotify’s editable Taste Profile—finally lets judgment lead. T...
chatbotsbehavingbadly.com
Looks finished. Isn’t.
The article is about the office villain with perfect margins and zero meaning: AI-generated workslop.
Read the article: chatbotsbehavingbadly.com/the-polished...
Board slides that say “AI reduces time-to-answer” should include the footnote: “...and raises error risk when we force brevity.”
Read the full article.
chatbotsbehavingbadly.com/glue-on-pizz...
AI has gone from novelty wingman to built-in infrastructure for modern dating—photo pickers, message nudges, even bots that “meet” your match before you do.
New episode is out now!
chatbotsbehavingbadly.com?p=S2E6
Everyone wants a piece of AI. Few know where to dig.
We live in the quiet layers beneath the hype — where signal hides, founders build, and advantage still compounds in silence.
That’s our work. That’s Access. Rights. Scale.™

medium.com/@markus_brin...
What happens when you let an AI run a real store for a month? Anthropic found out the hard (and hilarious) way.
Read the full story.
chatbotsbehavingbadly.com/the-comedy-o...
Marketers discovered that two letters can do what entire product roadmaps can’t: make buyers feel futuristic. Like stabilizers in processed food, “with AI” keeps the brand story glossy, prevents separation, and rescues the flavor when the recipe’s thin. chatbotsbehavingbadly.com/with-ai-is-t...
'With AI' is the new 'Gluten-Free'
Marketers discovered that two letters can do what entire product roadmaps can’t: make buyers feel futuristic. Like stabilizers in processed food, “with AI” keeps the brand story glossy, prevents separ...
chatbotsbehavingbadly.com
Online, we meet the idea of a person. AI supercharges that idea. The first date is where reality negotiates the contract. Here’s a human-first guide to using AI without catfishing yourself—or anyone else.
Your read for the weekend.

chatbotsbehavingbadly.com/the-chat-was...
New Stanford research shows “AI therapists” can reinforce stigma and mishandle crisis cues—sometimes answering lethal prompts like trivia. The fix isn’t a bigger model; it’s a different role: assistants for clinicians, not replacements.

chatbotsbehavingbadly.com/therapy-with...
Therapy Without a Pulse
New Stanford research shows “AI therapists” can reinforce stigma and mishandle crisis cues—sometimes answering lethal prompts like trivia. The fix isn’t a bigger model; it’s a different role: assistan...
chatbotsbehavingbadly.com
Today’s episode of Chatbots Behaving Badly is about why AI-generated images look textbook-ready and still teach the wrong history.

S2E4: "Pictures That Lie" chatbotsbehavingbadly.com?p=S2E4

Based on the article "Pictures That Lie" by Markus Brinsa.
chatbotsbehavingbadly.com/pictures-tha...
Classrooms are quietly outsourcing history to probability machines. My new Chatbots Behaving Badly article explains why these models fail at facts, how that misleads students, and what responsible use actually looks like. chatbotsbehavingbadly.com/pictures-tha...
Pictures That Lie
Classrooms are quietly outsourcing history to probability machines. A recent teacher training deck asked ChatGPT to “create a multimedia presentation on the Mexican Revolution.” The result looked text...
chatbotsbehavingbadly.com
AI isn’t just changing how we work. It’s changing how we think—often in ways we don’t notice until it’s too late.
This isn’t a glitch. It’s a warning.

Watch the carousel video. Then read the full piece here:

chatbotsbehavingbadly.com/the-illusion...
What happens when a chatbot tells you your darkest thoughts are true?

We’re finding out — and it’s not pretty.

chatbotsbehavingbadly.com/delusions-as...