Jyu'LogJun Yu: research, products, and writing.https://junyu.blog/enWeak-to-Strong Exploitation: Weak LLMs Orchestrating Aligned Models Toward Harmful Goalshttps://junyu.blog/posts/unaware/https://junyu.blog/posts/unaware/Weak LLMs as malicious orchestrators: a threat model, three mechanisms, causal evaluation, and future research on RL.Tue, 01 Jul 2025 00:00:00 GMTStable Signer: Balancing Three Stakeholders’ Interests in Image Provenancehttps://junyu.blog/posts/stable-signer/https://junyu.blog/posts/stable-signer/A message can change a generative model's decoder. The harder question is who chooses the change, who can read it, and who gets to make a claim from it.Sun, 01 Sep 2024 00:00:00 GMT