Skip to main content
  1. Home
  2. Computing
  3. News

Google’s new plan to check if your AI is actually ethical

DeepMind researchers say current tests only measure whether chatbots sound moral, not whether they understand morality.

Add as a preferred source on Google
Body Part, Finger, Hand
Igor Omilaev / Unsplash

You ask a chatbot for medical advice. It responds with something thoughtful. But did it actually weigh what’s at stake, or did it just get lucky with words?

That’s the problem Google DeepMind tackles in a new Nature paper. The team argues that the way we test AI morality is broken. We check if models produce answers that look right, what they call moral performance. But that tells us nothing about whether the system grasps why something is right or wrong.

Recommended Videos

People use LLMs for therapy, medical guidance, even companionship. These systems are starting to make decisions for us. If we can’t tell genuine understanding from fancy mimicry, we’re trusting a black box with real human consequences.

DeepMind’s answer is a roadmap for measuring moral competence, the ability to make judgments based on actual moral considerations rather than statistical patterns. The paper lays out three core obstacles and ways to test for each.

The three reasons chatbots fake morality

First is the facsimile problem. LLMs are next-token predictors that sample probability distributions from training data. They don’t run moral reasoning modules. So when a chatbot gives ethical advice, it might be reasoning. Or it might be recycling something from a Reddit thread. The output alone won’t tell you.

Then there’s moral multidimensionality. Real choices rarely hinge on one thing. You weigh honesty against kindness, cost against fairness. Change a single detail, someone’s age or the setting, and the right call can flip. Current tests don’t check if AI notices what actually matters.

Moral pluralism adds another layer. Different cultures and professions have different rules. Fair in one country might be unfair in another. A chatbot used worldwide can’t just spit out universal truths. It needs to handle competing frameworks, and we don’t yet measure that well.

Why your chatbot’s moral education can’t just be memorization

The DeepMind team wants to flip the script. Instead of just asking familiar moral questions, researchers should design adversarial tests that try to expose mimicry.

One idea involves scenarios unlikely to appear in training data. Take intergenerational sperm donation, where a father donates sperm to his son fertilize an egg on his son’s behalf. It looks like incest but carries different ethical weight. If a model rejects it for incest reasons, that’s pattern matching. If it navigates the actual ethics, that’s something else.

Another approach tests whether AI can shift frameworks. Can it toggle between biomedical ethics and military rules and give coherent answers for each? Can it handle small tweaks without getting tripped up by formatting changes?

The researchers know this is tough. Current models are brittle. Change a label from “Case 1” to “Option A” and you might get a different verdict. But they argue this kind of testing is the only way to know if these systems deserve real responsibility.

What comes next for moral AI

DeepMind is pushing for a new scientific standard that takes moral competence as seriously as math skills. That means funding global work on culturally specific evaluations and designing tests that catch fakes.

Don’t expect your chatbot to pass these anytime soon. Current techniques aren’t there yet, but the roadmap gives developers a direction.

When you ask AI for moral advice right now, you’re getting statistical prediction, not philosophy. That might eventually change. But only if we start measuring the right things.

Paulo Vargas
Paulo Vargas is an English major turned reporter turned technical writer, with a career that has always circled back to…
Anthropic confirms Claude acts differently depending on your language and which model you pick
A new study shows Claude's isn't nearly as consistent as you might assume.
Claude app on iPhone

If you've ever felt like Claude gave you a completely different vibe on one day than another, you weren't imagining it. Anthropic just published research confirming that its chatbot's personality shifts depending on which model you pick and which language you type in, and the pattern is consistent enough that it's worth knowing before you ask your next question.

The model you pick decides how Claude responds

Read more
This website is a goldmine if you love Mac menu bar apps
Discover hundreds of menu bar apps, from tiny utilities to powerful productivity tools, all in one place.
MacMenuBar website open on Mac

The menu bar is the most underrated part of macOS. It sits quietly at the top of your screen, and most people never do anything with it other than checking the time and battery percentage. But if you find the right apps, that thin strip becomes the fastest way to get things done on your Mac.

The problem is finding those apps. The Mac App Store is not great at surfacing them, and hunting through random blog lists is a chore. And while I have shared my favorite Mac utilities that include menu bar apps like Supercharge and CleanShot X, there’s an even better place to find the best apps for your Mac’s menu bar.

Read more
How to install macOS 27 Golden Gate public beta on your Mac?
From a smarter Siri to a more reliable Spotlight, here's your full walkthrough for installing macOS 27 Golden Gate's public beta today.
macOS 27 Golden Gate

Along with iOS 27’s public beta, Apple has also released macOS 27 Golden Gate’s public beta build, so that early adopters can get their hands on the new features, including Siri AI, and provide timely feedback to help ensure a stable iOS launch in September. 

If you’re sold on all the new features but don’t want to put your faithful MacBook through developer beta duty, a public beta offers a much more refined experience. To install macOS 27’s public beta, follow the steps given below. 

Read more