Flawed AI, Risky Deployments
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Thursday, June 18, 2026
Here's something most people won't piece together from today's headlines: the UK is about to let flawed facial recognition decide the fate of asylum-seeking children, Midjourney just quietly stepped into medical imaging, and Google is already preparing for AI agents that go rogue. Three very different stories, one uncomfortable truth - we're deploying AI into life-or-death situations faster than we're figuring out how to keep it honest.
Today in AI:
- The UK's AI Age-Guesser Gets Kids Wrong - The British government plans to use facial age estimation on asylum seekers despite internal tests showing the tech regularly mistakes children for adults and contains serious bias problems. It's moving forward anyway. Wired
- Midjourney Enters the Exam Room - The AI image company best known for fantasy art announced Midjourney Medical, a push into medical technology. Details are thin, but the ambition is clear: from generating portraits to generating diagnostic insights. Midjourney
- Google DeepMind Preps for Rogue AI Agents - Google is borrowing cybersecurity tactics to build safeguards for autonomous AI agents before they go off-script. Translation: they're building the guardrails while the car is already on the highway. Axios
- Yann LeCun Calls xAI a 'Failure' - The AI pioneer told CNBC that Musk's xAI can't retain top talent, won't compete on the frontier, and is mostly renting out its data centers to recoup costs. xAI's AI segment posted a $2.5 billion operating loss last quarter. CNBC
- How to Evict Gemini From Your Google Docs - A TechCrunch writer got so annoyed by Google's AI pop-ups that they wrote a full guide on removing them instead of doing actual work. The kicker: when asked for help removing itself, Gemini tried to dodge the question. TechCrunch
- Senator Hawley Rips AI 'Cheerleaders' at Teamsters Convention - The Republican senator blasted AI giants and backed union priorities at a closed-door Teamsters event, signaling that AI's labor impact is becoming a bipartisan political issue. Axios
- Flagright Raises $12.5M to Chase Financial Crime With AI - The YC-backed, Berlin-based startup sells AI-powered compliance tools to banks and fintechs across 35-plus countries. The Series A will fund US expansion as legacy compliance systems get replaced. Sifted
- Pixi Sends AI Characters Through Your Texts - A new iOS app from a former DreamWorks and Apple designer lets you send AR characters via iMessage that react to real-world surroundings using on-device AI. Your birthday texts just got weirder. TechCrunch
Today's Takeaway:
Look at today's biggest stories through a single lens: accountability gaps. The UK's Home Office tested facial age estimation technology, found it mistakes children for adults and shows serious bias - particularly against the largest demographic of asylum seekers subject to age checks - and decided to deploy it anyway. As Wired reports, children incorrectly classified as adults can lose legal protections and end up in adult detention. The internal government report documenting these failures exists. The rollout proceeds regardless.
Now hold that next to Google DeepMind building safeguards for autonomous AI agents, per Axios. Here's the thing: one company is preemptively trying to control AI before it causes harm, while a government is knowingly deploying flawed AI into decisions that alter human lives. The difference isn't technical - it's institutional. Companies building AI have started to realize that "move fast and break things" gets expensive when regulators come calling. Governments, ironically, face less market pressure to get it right. The people most affected by flawed AI - asylum seekers, gig workers, the unbanked - are consistently the people with the least power to push back. That's the pattern worth watching: AI's worst deployments don't happen where oversight is strongest. They happen where the affected people have the weakest voice.
💡 Fluency Moment - Building your AI fluency, one term at a time.
"Guardrails"
In plain English: Safety rules built into AI to stop it from doing harmful or unintended things. Think of it like: Like bumpers in bowling - they do not guarantee a strike, but stop the worst outcomes. Why you'll hear about it: Google is literally building these now for rogue AI agents making real-world decisions.
The Bottom Line
The Pattern: AI is spreading into high-stakes territory - immigration decisions, medical imaging, autonomous agents, financial compliance - and the speed of deployment is outpacing the speed of accountability. The organizations with the most power to cause harm often face the least friction in doing so.
Why It Matters: When AI screws up a birthday text, you laugh. When it screws up a child's age at a border, that child ends up in an adult detention facility. The stakes are wildly uneven across these use cases, but the underlying technology and its known flaws are remarkably similar. Your ability to spot where AI belongs and where it doesn't is becoming a core literacy skill.
Your Move: Next time someone pitches you an AI tool - for your business, your workflow, your hiring process - ask one question before anything else: what happens when it's wrong, and who bears the cost? If the answer is "someone with no power to complain," that's your red flag.
What We're Working On
✨ Founding Cohort Special - 60% Off! - Use code MAF20 to join for just $20/month (regularly $50). Get weekly group sessions & workshops, self-paced courses for all levels, access to tools & templates, challenges with peer feedback, and 24/7 support community. → Join Now
✨ Free 30-Minute AI Consultation - Discover how My AI Fluency can help your business unlock the potential of AI. We'll discuss your goals, explore practical AI opportunities for your industry, and outline clear next steps. → Schedule Free Call
✨ How AI-Fluent Are You? - Test your AI fluency with our interactive quiz. See how you stack up and discover what to learn next. → Take the Quiz
💬 Community | 📞 Book a Consultation | 🌐 Website
Fluently yours, The My AI Fluency Team