Name: Urgent research needed to tackle AI threats, says Google AI boss | BBC News
Uploaded: 2026-02-28T12:02:12.000Z
Duration: 3 min 17 s
Description: Thousands of YouTube videos with English-Chinese subtitles! Now you can learn to understand native speakers, expand your vocabulary, and improve your pronunciation...

More research on the threats of artificial intelligence needs to be done urgently.

That's what the boss of Google DeepMind has told BBC News.

Speaking to our technology editor, Zoe Kleinman at the AI Impact Summit in Delhi, Sedemis Hassabis said the industry wanted smart regulation for the real risks posed by the technology.

Of course, there's all the amazing benefits that we work towards, like better health care, better drugs, helping with climate change and energy.

I think all of these things are actually on the cusp of happening.

One is bad actors repurposing these technologies for harmful ends.

And then the second worry is a more technical risk, which is as these AI systems get more powerful, more autonomous, maybe entering the agentic era, how do we make sure we can build robust enough guardrails to keep them doing what we want them to do?

I think there's still an open research question and I think more research needs to be done urgently.

I think there are many ways I think it could work, but

Do you worry about how best to prepare your children for what's coming down the track?

I think there's always a worry about that with the next generation.

I mean, I've seen it myself when I grow up with computers and.

you see it here in India, they're actually really excited, I think, about the opportunities and that these new tools will afford them.

So I think part of it is just making sure they're equipped with the right skills, maybe different skills than before, to make use of these incredible new tools.

And I think the next 10 years will be almost like having superpowers in terms of what you're able to produce and do.

You know, it used to be computer science and now I can code.

Perhaps it's prompt engineering, but will there come a time when I will prompt itself?

Yeah, well, I think it's still very important to have a STEM education.

I would still say, like, if you know how to code or do math.

Even though these systems will start being able to code, you still need to orchestrate them and actually understand what it is they're doing and give them the problem that they need to solve.

If you have a technical background, I think you'll still be an advantage in using these systems.

But I think the exciting thing is that it's.

If these systems are able to code, going to open up the space for creatives to actually make and build many more things and prototype and iterate more quickly.

So I think he'll kind of expand who's able to create and then maybe the.

The key thing becomes taste and creativity and judgment.

The last time we spoke, you said you didn't want to move fast and break things, but

it feels a bit like we are moving fast and we are breaking things, aren't we?

Well, I think we're doing the best we can as we're only one player in the ecosystem.

We're an important player and I think we try to be a really good role model with being both.

You know, we like to talk about being both bold and responsible.

So bold with the opportunities, things like Alpha Fold, curing diseases.

These amazing things that we can do with these tools.

And the reason, you know, I started many, many years ago now working on AI, but

we've also got to be thoughtful, as thoughtful as possible as we deploy these very powerful systems out into the world and we try to get that balance right and try to act as a role model for doing that.

We don't always get things right, but I think we get it more correct than most.