Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • COMMENT

We need a new ethics for a world of AI agents

The deployment of capable AI agents raises fresh questions about safety, human–machine relationships and social coordination.
By
  1. Iason Gabriel

    1. Iason Gabriel is a senior staff research scientist at Google DeepMind in London.

  2. Geoff Keeling

    1. Geoff Keeling is a staff research scientist in Google’s Paradigms of Intelligence Team in London.

  3. Arianna Manzini

    1. Arianna Manzini is a senior research scientist at Google DeepMind in London.

  4. James Evans

    1. James Evans is a visiting faculty member in Google’s Paradigms of Intelligence Team in Chicago, Illinois, and a professor at the University of Chicago and the Santa Fe Institute in New Mexico.

Illustration of a series of floating panels with images of flight results, AI neural networks and code with a few of them showing error messages. Behind the panels is a human face.

Illustration: Ibrahim Arafath

Artificial intelligence (AI) developers are shifting their focus to building agents that can operate independently, with little human intervention. To be an agent is to have the ability to perceive and act on an environment in a goal-directed and autonomous way1. For example, a digital agent could be programmed to browse the web and make online purchases on behalf of a user — comparing prices, selecting items and completing checkouts. A robot with arms could be an agent if it could pick up objects, open doors or assemble parts without being told how to do each step.

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Learn more

Prices may be subject to local taxes which are calculated during checkout

Nature 644, 38-40 (2025)

doi: https://doi.org/10.1038/d41586-025-02454-5

References

  1. Russell, S. & Norvig, P. Artificial Intelligence: A Modern Approach (Pearson, 2020).

    Google Scholar 

  2. Russell, S. Human Compatible: Artificial Intelligence and the Problem of Control (Penguin, 2019).

    Google Scholar 

  3. Lu, C. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2408.06292 (2024).

  4. Sharkey, L. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2501.16496 (2025).

  5. Kirk, H. R., Gabriel, I., Summerfield, C., Vidgen, B. & Hale, S. A. Hum. Soc. Sci. Commun. 12, 728 (2025).

    Article  Google Scholar 

  6. Shanahan, M., McDonell, K. & Reynolds, L. Nature 623, 493–498 (2023).

    Article  PubMed  Google Scholar 

  7. Manzini, A. et al. In Proc. AAAI/ACM Conf. AI Ethics Soc. 7, 943–957 (2024).

    Article  Google Scholar 

Download references

Competing Interests

J.E. maintains a commercial relationship with Google.

Subjects

Latest on:

Nature Careers

Jobs

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing

Search

Advanced search

Quick links