Professor Philip Torr, Oxford University

This seems a bit like punishing the victims rather than perpetrators

Dinner to celebrate Oxford's Deep Mind scholars with Deep Mind founder, Demis Hassabis, and new incoming PhD, Anda! Thanks to Deep Mind and Demis for generously supporting so many students.

In the beautiful Divinity school of Oxford.

New Ellison institute at Oxford, set to bring, 100 or 1000s of jobs, lnkd.in/e6HA-_ZP

uper happy to be one of the organizers of this, eurips.cc, now an option to officially present your papers in Europe and save greenhouse gas!!!!! Please repost and spread the word!!! Please repost!!!!

Co-authored with: Isaac Friend, Keir Reid, Igor Krawczuk, Vincent Wang, @jakobmokander.bsky.social kander.bsky.social , @philiptorr.bsky.social , Julia C Morse and Robert Trager

As AI proliferates I am also interested in how we can prevent some Orwellian world coming about-it's a societal issue at its heart-but here is a technical paper in the mean time. | Philip Torr, FR... As AI proliferates I am also interested in how we can prevent some Orwellian world coming about-it's a societal issue at its heart-but here is a technical paper in the mean time.

As AI proliferates I am also interested in how we can prevent some Orwellian world coming about-it's a societal issue at its heart-but here is a technical paper in the mean time. www.linkedin.com/posts/philip...

YouTube video by Sabine Hossenfelder AI is becoming dangerous. Are we ready?

Youtuber Sabine Hossenfelder has picked up on our paper on how to fool AI agents, you can watch her desrcibing our work here: www.youtube.com/watch?v=KY7_....

Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers

Wei Pang, Kevin Qinghong Lin, @edwardjian.bsky.social, Xi He, @philiptorr.bsky.social

tl;dr: great stuff!

arxiv.org/abs/2505.21497

Had the privilege of speaking at Oxford Brookes graduation today, was really fun event, and inspiring to see. Amazing to see people off on their, hopefully, exciting journeys.

Meet virtually with Cisco Webex. Anytime, anywhere, on any device. Simple, modern video meetings for everyone on the world's most popular and trusted collaboration platform.

To join us on the day go to: dit.webex.com/dit/j.php?MT...
Password: trade123 (87233123 when dialling from a phone or video system)

#policy #universities #tech #researchers #professionals #ai | Polly Benjamin 📢 Are you an AI researcher or engineer looking for your next career move? Join leading UK AI business professionals and academics on a webinar on 14th May to explore the strength of the UK’s AI sector...

Just about to speak at this to promote UK PLC! www.linkedin.com/posts/polly-...

Job Details | Philip Torr JOB!!!! Adel is amazing

research position www.linkedin.com/posts/philip...

Shh, don't say that! Domain Certification in LLMs Domain Certification - A novel framework providing provable, adversarial defenses for LLMs safety.

Read more: cemde.github.io/Domain-Certi...

Thanks to my amazing collaborators:
- @alasdair-p.bsky.social, Preetham Arvind, @maximek3.bsky.social, Tom Rainforth, @philiptorr.bsky.social, @adelbibi.bsky.social at @ox.ac.uk
- Bernard Ghanem at KAUST
- Thomas Lukasiewicz at @tuwien.at.

(7/7)

Cooperation on AI safety is necessary but also comes with potential risks. In our new paper, we identify technical AI safety areas that present comparatively lower security concerns, making them more suitable for international cooperation—even between geopolitical rivals. lnkd.in/e_FzCFW9

Encode Project Q&A – AI for Drug Discovery · Zoom · Luma Join us for a virtual info session to learn more about the Encode: AI for Science Fellowship. 🧪 This session will feature Philip Torr, one of of our faculty…

We will be hosting a live Q&A to discuss our project on April 10 at 11:30AM BST.

Join here: lu.ma/encode-proje...
More about the Encode: AI for Science Fellowship: encode.pillar.vc/blog-posts/f...

Papers with Code - @Bench: Benchmarking Vision-Language Models for Human-centered Assistive Technology No code available yet.

I wanted to coin the "Torr Law", which is that if you think of a research idea, it will magically appear on the internet within a few days, case in point, this is a cool first step on the way to making AI scientists paperswithcode.com/paper/bench-...

Strengthening AI Resilience | AISI Work | Philip Torr Excited to be work with the UK AI Security Institute, even more important than normal in these turbulent times.

Excited to be working with the UK Excited to be work with the UK AI Security Institute, even more important than normal in these turbulent times.

www.linkedin.com/posts/philip...

Do Sparse Autoencoders Generalize? A Case Study of Answerability Sparse autoencoders (SAEs) have emerged as a promising approach in language model interpretability, offering unsupervised extraction of sparse features. For interpretability methods to succeed, they m...

📄 Dive In:

paperhttps://arxiv.org/abs/2502.19964.

Work led by
@lovisheindrich.bsky.social
- [Lovis is in the job market, you should hire him, he is great!]

and thanks to
@philiptorr.bsky.social and @vthost.bsky.social
for advising.

Attacking Multimodal OS Agents with Malicious Image Patches Recent advances in operating system (OS) agents enable vision-language models to interact directly with the graphical user interface of an OS. These multimodal OS agents autonomously perform computer-...

🏛️ This work was made possible with OATML and TVG at the University of Oxford (@ox.ac.uk). Special thanks to @yaringal.bsky.social, @adelbibi.bsky.social, @philiptorr.bsky.social, and @alasdair-p.bsky.social for their contributions.

📖 Read the paper: www.arxiv.org/abs/2503.10809

Stable Virtual Camera: Generative View Synthesis with Diffusion Models

Jensen (Jinghao)Zhou, Hang Gao, Vikram Voleti, @adyaman.bsky.social, Chun-Han Yao, @markboss.bsky.social, @philiptorr.bsky.social, Christian Rupprecht, Varun Jampani

arxiv.org/abs/2503.14489

NeurIPS participation in Europe | Philip Torr | 10 comments this is a great idea, having a parallel Neurips in Europe is long over due, especially as travel to the US is difficult now for a lot of scientists. Thoughts?… | 10 comments on LinkedIn

this is a great idea, having a parallel Neurips in Europe is long over due, especially as travel to the US is difficult now for a lot of scientists. Thoughts?

www.linkedin.com/posts/philip...

Reposting this opportunity as it is a great one! | Philip Torr Reposting this opportunity as it is a great one!

This is an amazing opportunity, please repost

We have two visiting posts open in collaboration with Pillar,VC. You'll get:
- £115k salary
- £100k compute
- Access to AI models, tools, and agent systems, pls repost

www.linkedin.com/posts/philip...

i fundamentally dont agree, if you resign then you dont have any influence, its a bit like not voting in an election, which i also dont agree with.

long time no physical see, hope you are coping ok with it

Hit home today that Elon’s zooming in to a rally of the German hard right to tell them to stop feeling bad about the past and embrace German glory was right before Holocaust Remembrance day.

Note Oxford has advertized many AI positions including a new faculty position, lnkd.in/eGYekxGy, see more here eng.ox.ac.uk/jobs/

UK ‘second most attractive country for investment’, survey finds The survey of chief executives, published ahead of the World Economic Forum, follows an upgraded growth forecast from the IMF

www.theguardian.com/business/202...

super excited to be speaking at starmus this year www.starmus.com

Why ‘Beating China’ in AI Brings Its Own Risks The US is increasingly intent on winning the AI race with China. Experts say this ignores the benefits of collaboration—and the danger of unintended consequences.

www.wired.com/story/why-be...