Imagine bringing home a report card and finding out that out of nine students in your class, the best score was a C+. That’s not a hypothetical school assignment — it’s exactly what just happened to the world’s most powerful AI companies.
What just happened
On July 7, 2026, an independent research group called the Future of Life Institute released its newest AI Safety Index — think of it as a report card, but instead of grading students, it grades AI companies. The researchers looked at nine major labs — Anthropic, OpenAI, Google DeepMind, Meta, xAI, DeepSeek, Mistral, Z.ai, and Alibaba Cloud — and scored each one on how carefully and responsibly they build and test their AI systems.
The results were humbling. Anthropic came out on top with a C+. OpenAI and Google DeepMind each earned a C. Meta scored a D+. Several companies, including xAI, DeepSeek, and Mistral, failed outright with an F. Not a single company earned an A or a B. Researchers also flagged a worrying trend: some companies that had previously promised not to use AI for military purposes have quietly reversed that promise.
A report card you already understand
You already know how report cards work. A good grade usually means the work was checked carefully, mistakes were caught early, and the assignment was taken seriously. A low grade often means corners were cut, or nobody double-checked the work before handing it in.
AI safety grades work the same way. They measure whether a company tests its AI carefully before releasing it to millions of people, whether it’s honest about the risks, and whether it has a real plan for what happens if something goes wrong. A company can build a brilliant, powerful AI and still earn a low grade — because being smart and being safe are two different report cards entirely.
Why this matters to your family
These AI systems already help students write essays, help doctors read scans, help small businesses answer customer emails, and help parents plan a week’s meals. If the companies building them are mostly earning C’s and D’s on safety, that affects everyone who uses the technology — including your household. A “smart” tool is not automatically the same as a safe or trustworthy one, and it’s worth teaching our kids to notice the difference.
I often remind my students not to become dependent on AI, but to treat it as a tool. AI can help us generate ideas, explain difficult concepts, organise information, and complete certain tasks faster. However, after using many AI tools myself, I have learned that they are still far from perfect. Their answers may sound confident while containing mistakes, missing important context, or failing to understand the real purpose behind a task.
This is why human involvement remains essential. We must question the output, verify the facts, correct errors, and decide whether the answer is ethical, useful, and appropriate. AI can assist our thinking, but it should not replace our judgment. It can produce a starting point, but humans must provide experience, creativity, responsibility, and wisdom.
I want my students—and my own children—to become capable users of AI, not passive followers of it. The goal is not merely to finish work faster. The goal is to use AI to improve how we learn, create, and solve problems while continuing to develop our own abilities. If we allow AI to do all our thinking, convenience may slowly weaken our competence. But when we use it with discipline, AI can strengthen what humans are already capable of doing.
Questions worth thinking about
- If your favorite AI tool got a C on its safety report card, would you still trust it with your homework, or your personal information?
- Why do you think a company might promise not to use AI for one purpose, and then quietly change its mind later?
- What would an A+ AI safety report card need to include?
- Who should be responsible for grading AI companies — governments, independent researchers, or the companies themselves?
The lesson underneath the headline
Being capable and being safe are not the same skill. The most powerful AI in the world is still only as trustworthy as the people building it and the choices they make when nobody’s watching. That’s true of AI. It’s true of businesses. It’s true of us.
A report card isn’t the end of the story — it’s a chance to improve before the next one comes out. As AI keeps getting smarter, the real question worth sitting with is simpler: are the people building it getting wiser? That’s a grade worth watching, right alongside the technology itself.
Keep exploring
- How I Use Claude, ChatGPT, Gemini, and Copilot Differently
- Your Computer Just Got a New Assistant: What ChatGPT Work Means for Students, Teachers, and Small Business Owners
- The Robot That Learns Like a Student, Not by Being Reprogrammed
If this made you think differently about the AI tools in your home or classroom, share it with a student, teacher, or parent who’s curious about using AI wisely — and subscribe so you don’t miss the next one.
JOSIAH Miri 