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

Customizable AI systems that anyone can adapt bring big opportunities — and even bigger risks

Open and adaptable artificial-intelligence models are crucial for scientific progress, but robust safeguards against their misuse are still nascent.
By
  1. Yarin Gal

    1. Yarin Gal is a research director at the UK AI Security Institute in London, UK.

  2. Stephen Casper

    1. Stephen Casper is a PhD student at Massachusetts Institute of Technology, Cambridge, USA, and was previously a researcher at the UK AI Security Institute in London, UK.

Two people speak two each other at a conference table onstage in front of a blue screen.

Meetings such as the AI Seoul Summit in 2024 aim to ensure responsible AI development. Credit: Zoe-Rose Herbert/DSIT (CC BY 2.0)

In the past three months, several state-of-the-art AI systems have been released with open weights, meaning their core parameters can be downloaded and customized by anyone. Examples include reasoning models such as Kimi-K2-Instruct from technology company Moonshot AI in Beijing, GLM-4.5 by Z.ai, also in Beijing, and gpt-oss by the California firm OpenAI in San Francisco. Early evaluations suggest that these are the most advanced open-weight systems so far, approaching the performance of today’s leading closed models.

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 646, 286-287 (2025)

doi: https://doi.org/10.1038/d41586-025-03228-9

References

  1. Internet Watch Foundation. What has Changed in the AI CSAM Landscape? (IWF, 2024).

  2. AI Action Summit. International AI Safety Report (UK Government, 2025).

  3. O’Brien, K. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2508.06601 (2025).

  4. Qi, X. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2310.03693 (2023).

  5. Qi, X. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2412.07097 (2024).

  6. Che, Z. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2502.05209 (2025).

Download references

Competing Interests

The authors declare no competing interests.

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