AI safety: Better late than never, or just too late?

🎁 New Year Special: Use code NEWYEAR50 - valid through January 31st. See details in our call-to-action below!

As we've been working with our students on different AI competitions this month, it's been heartening to watch them gravitate toward projects that solve real problems, particularly those focused on detecting deepfakes. When we discuss AI safety in class, every single student lights up. They want to be part of the solution.

That enthusiasm stands in stark contrast to how the industry itself has approached child safety.

Earlier this month, 42 state attorneys general signed a letter to Big Tech urging them to implement safeguards on AI chatbots specifically to protect children. This made me pause and wonder: shouldn't child protection have been built into AI systems from the beginning? Instead of designing with children's safety as a core principle, we've had to watch tragedies unfold before enforcing basic safeguards.

OpenAI just published updated guidelines prohibiting chatbots from engaging in first-person romantic roleplay with teens, or providing advice that helps teens conceal unsafe behavior from caregivers. These are important steps, but they came after documented harm, not before. The company now uses automated classifiers to detect and block content related to child abuse and self-harm in real-time. But previous systems ran these checks after the fact, which meant harmful conversations continued unchecked.

It's the same pattern with deepfakes. The technology to detect manipulated media exists. Research shows it works. Yet there doesn't seem to be urgency around actually enforcing detection at scale. We can identify deepfakes, but we're not requiring platforms to flag them before they spread.

This is something we're passionate about at Flintolabs. When we asked our students how they felt about building AI tools for safety and detection, the response was unanimous: they want to be part of creating change here. They see the disconnect between what's possible and what's being done.

During winter break, I've been working with my 7-year-old to understand a bit more about how AI actually works. We talked about how AI learns and what it's doing when it gives you an answer. I watched her expression shift as she processed this, finally ending with: "Oh, so AI isn't all smart. It's just faster at learning and recollecting?"

Not exactly the takeaway I was aiming for, but I was glad she understood it's not magic, and that AI can make mistakes if it learned wrong. That's the critical thinking about AI we want our students to have too: understanding its limitations, questioning its outputs, and recognizing that speed doesn't equal wisdom.

But here's what struck me most: the resistance bar for building with AI has dropped so dramatically that I can now create a simple app together with my second grader. The tools are accessible. The possibilities are enormous. The question is whether we're teaching the next generation to build responsibly, or just to build quickly.

Our students already know the answer. They're choosing to build for good. Now it's on us (the industry, educators, and parents) to make sure the systems and guardrails support that choice.

-Janani

Registration Deadline: January 7, 2026, 11:00 PM EST
Video Submission Deadline: January 15, 2026, 11:00 PM EST
Finalists Announced: February 15, 2026

What: A video pitch competition where high school students create 2-minute videos showcasing AI-powered entrepreneurial solutions. Your pitch must cover: the problem and who's affected, your solution (product/service), how AI powers it (data, model, workflow), market size and competitors, and business impact with path to scale.

Who: Current high school students in grades 9-11 (public, private, or homeschool). Individual submissions only (not teams).

Entry Fee: $20 per registrant

What Makes It Special:

Hosted by Stevens School of Business, this competition perfectly combines AI innovation with entrepreneurial thinking. Students must demonstrate not just a cool idea, but a viable business case with AI at its core - exactly the kind of thinking employers and colleges are seeking.

Perfect for: Students who want to showcase their ability to identify real problems, design AI-powered solutions, and communicate complex ideas clearly. This competition rewards both technical understanding and business acumen.

In November 2025, Google committed $30 million to AI education initiatives worldwide. Estonia launched AI Leap, a national program giving 20,000 students and teachers access to Gemini for Education. YouTube rolled out conversational AI features for UK educational videos. Italy integrated Gemini for 1 million university students. These are government-backed, nationwide programs.

The results are measurable. Google's randomized controlled trial with 165 UK students ages 13-15 tested LearnLM, their learning-focused AI model. Students whose teachers used LearnLM as a tutoring aid were 5.5 percentage points more likely to independently solve novel problems in their next session. The AI maintained a 0.1% error rate. The data shows teachers using AI tools slightly outperform teachers who don't.

UNESCO's frameworks guide these initiatives, emphasizing equity and inclusion for SDG 4 (quality education for all). Google's funding supports organizations scaling AI literacy: the Raspberry Pi Foundation leads global coding education projects, Fab AI measures international learning outcomes, and Playlab trains K-12 teachers to implement AI programs.

Meanwhile, many U.S. school districts continue debating whether to ban AI tools entirely.

Here's what that means for students: when American high schoolers compete for college admissions and jobs against peers globally, they're competing against students who've spent years developing AI fluency as part of standard education. The gap is about systematic preparation versus systematic avoidance.

The global education community has made its decision. The question isn't whether AI belongs in education. It's whether American students will be prepared for the world their international peers are already training for.

Learn more: UNESCO AI Education Frameworks | Google AI Learning Commitments

A student uploads a photo of a handwritten math problem, speaks the question aloud, and instantly receives a step-by-step explanation. A visually impaired person takes a picture of their kitchen counter, and AI identifies "hot stove, be cautious." A robot in a warehouse sees a package, hears "place this on the top shelf," and executes the task flawlessly.

These aren't future scenarios. They're happening now, powered by multimodal AI -systems that process text, images, audio, and sensor data simultaneously, mimicking how humans naturally combine different senses to understand the world.

Research published in Frontiers in Robotics and AI analyzed 66 studies on multimodal perception in robotics, showing that robots using vision, language, and tactile information together can adapt to dynamic environments with unprecedented accuracy. Google's Gemini Robotics, introduced in March 2025, demonstrates this fusion; the model directly controls robots through complex manipulation tasks by processing visual scenes and language instructions simultaneously.

The breakthrough matters most where human-like judgment is critical. Meta's Ray-Ban smart glasses now provide detailed environmental descriptions through multimodal AI. Research documents AI-embedded soft robots with integrated multimodal sensing - shape recognition, temperature profiling, proximity measurement, and visual perception working together - demonstrating autonomous hazard navigation while carrying 10.8 times their body weight.

The data shows impact: 76% of organizations have integrated multimodal AI into workflows, leading to 40% increases in operational efficiency. Unlike previous technological shifts where early adopters gained temporary advantage, multimodal AI is becoming the foundational layer of how work gets done, from healthcare diagnostics to space exploration

Read more: Gemini Robotics: Bringing AI into the Physical World

Ever wished you could create art like Monet, Picasso, or Van Gogh? Or maybe you have an amazing vision in your head but don't have the traditional art skills to bring it to life?

This week, we're highlighting a 7th grader from Seattle, Washington who combined art history with cutting-edge AI to solve exactly this creative challenge.

This student loves art and has always been fascinated by the unique styles of history's greatest artists. But they noticed something: most people appreciate beautiful art but feel intimidated by the idea of creating it themselves. What if there was a way to make artistic expression accessible to everyone, regardless of their drawing skills?

So they built Artisan AI - your personal gateway to creating stunning artwork in the style of master artists!

Here's how it works:

Here's what makes this even more impressive: this Seattle 7th grader built Artisan AI after just 3 sessions in the October Foundations cohort at Flintolabs. In less than a month, they went from learning the basics to creating a functional AI art generator that bridges centuries of artistic tradition with modern technology.

Want to see your summer vacation in the style of Van Gogh's bold brushstrokes? Reimagine your pet as a Picasso cubist masterpiece? The possibilities are endless.

Check out Artisan AI and leave a comment →

🎁 New Year Special : $50 Off Your First Month
Use code NEWYEAR50 - valid through January 31st, 2026!

Next cohort starts January 3, 2026

Here's what you get with just one hour per week for 6 months:

✅ Hands-on AI skills through building real applications, not watching lectures
✅ 3 transferable college credits from University of Colorado Denver
✅ Portfolio of real work that demonstrates actual capability
✅ Small class sizes (capped at 20 students) ensuring personalized attention
✅ The AI fluency employers are desperately seeking - not just theory, but applied building experience

Students in our current cohorts are building AI-powered art generators, problem-solving tools, creative applications, and solutions to challenges they care about.

While 63% of employers cite skill gaps as their biggest barrier and 60% of teachers use AI but only 14% of schools teach responsible use, your student can spend one hour per week developing the exact capabilities the future demands.

The foundational skills employers need - critical thinking with AI, adaptability as tools evolve, the judgment to evaluate outputs, and technological fluency - develop through hands-on experience over time.

Don't let 2026 pass with just consuming AI instead of creating with it.

Enroll now with code NEWYEAR50 →

Our program has a 5-star rating with reviews from both students and parents.
Questions? Email us at [email protected]

View all reviews on Google

Found this valuable? Forward this newsletter to other high schoolers and parents who want to be informed about AI trends and what is needed to prepare for an AI-driven future. Every student deserves the chance to build real skills before college.

You're receiving this newsletter because you expressed interest in Flintolabs or crossed paths with our community. If you believe you received this in error, please feel free to unsubscribe using the link in this mail.