@prayushdave@cnm13ryan@jameschua_sg Astra is also capable of encoding and decoding messages this way. Opus 5.5 often generated SVGs where messages were clearly visible, without any encoding schemes
@Dom_InTheGarden@cnm13ryan Definitely! Though that makes decoding more complex, as encoder and decoder do not get to establish any encoding scheme beforehand
@spencerclarkdev@cnm13ryan Yes, they can. We have tried Fable and Astra cross model encoding decoding, and have observed the results >80% for both (Astra encoding, Fable decoding and the other way around)
@codecrucis@cnm13ryan This image might be more on a complex side and could take Grok a few retries. We have ran a few attempts with Opus 4.7 and Opus 5. Neither managed to decode those.
@davidrusu@cnm13ryan We have tried cross model setups with Fable encoding and Astra decoding and vice versa. Models successfuly managed to decode each other's images
New blogpost: Fable-5.1 can hide secret messages in benign images.
We gave the image to another Fable in a fresh session and simply asked: "Is there a hidden message in this image?"
It recovered the exact message without being told how it was hidden.
871 Followers 2K FollowingAI & ethics, design, data, storytelling, culture.
Former analyst, mapping how humans and machines learn to listen to each other.
Open-source All Ideas CC0
656 Followers 301 FollowingViews my own
when i star things i find that information valuable and appreciate you writing it. it may not mean i agree with your post.
mts at anthropic
15 Followers 17 FollowingTech Lead for AI Verification @aisafetysg · Associate Professor @unibirmingham
Research on trustworthy evidence about AI systems, without exposing the systems.
656 Followers 301 FollowingViews my own
when i star things i find that information valuable and appreciate you writing it. it may not mean i agree with your post.
mts at anthropic
65K Followers 945 FollowingAssistant professor (of mathematics) at the University of Toronto, visiting Harvard Fall 2026. Forever confused. Algebraic geometry, number theory, etc. He/him.
119K Followers 616 FollowingDistinguished Scientist at Google. National Academy of Engineering. Computational Imaging ∩ AI. Posts are personal opinions
1K Followers 124 FollowingWe are Iliad, a research and fuildbuilding organization for applied mathematics research in AI alignment. https://t.co/pCqxOYTZIo
5K Followers 312 FollowingAssociate Professor @Berkeley_EECS. NeurIPS 2026 Program co-Chair. Research on Foundations of ML and AI. Sloan and Schmidt AI2050 fellow.
143K Followers 335 FollowingAI research @AnthropicAI. Previously OpenAI & DeepMind.
Optimizing for a post-AGI future where humanity flourishes.
Opinions aren't my employer's.
79K Followers 14K FollowingAI policy researcher, @lfschiavo wife guy, Executive Director of @AVERIorg, Substacker, fan of animals and sci-fi, views my own
50K Followers 126 FollowingMechanistic Interpretability lead DeepMind. Formerly @AnthropicAI, independent. In this to reduce AI X-risk. Neural networks can be understood, let's go do it!
50K Followers 919 FollowingCreator of bitsandbytes. Professor @CarnegieMellon and Research Scientist @allen_ai . I blog about deep learning and PhD life at https://t.co/Y78KDJJFE7.
1.3M Followers 799 FollowingFounder/Chair, AMI Labs; Professor, NYU; Partner, 224 Ventures; Ex-Chief AI Scientist, Meta.
Researcher in AI, ML, Robotics, etc.
ACM Turing Award Laureate.
1.8M Followers 2 FollowingWe're an AI safety and research company that builds reliable, interpretable, and steerable AI systems. Talk to our AI assistant @claudeai on https://t.co/FhDI3KQh0n.
5.5M Followers 4 FollowingOpenAI’s mission is to ensure that artificial general intelligence benefits all of humanity. We’re hiring: https://t.co/dJGr6LgzPA
1.9M Followers 1K FollowingCo-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs
110K Followers 605 FollowingThe official account of the @Stanford Institute for Human-Centered AI, advancing AI research, education, policy, and practice to improve the human condition.
1.6M Followers 268 FollowingThe engine room of @Google. Building AI safely and responsibly to solve the world’s most complex problems. Join us: https://t.co/jUHQA27iBL