
I love the smell of cryptanalysis in the afternoon!

CS, Cryptography, Cryptanalysis, Privacy, Biometrics, Computer Security. (personal account at @mkilmo). He/Him. Also @CryptoOrrDun@ioc.exchange

I love the smell of cryptanalysis in the afternoon!

I guess that Asiacrypt will have a panel on AI first breaking hash functions or public-key crypto? Or will the first panel at Asiacrypt will be about AI first stripping cryptographers of their jobs? So... let's argue which panel will be first!

I am grading the crypto exams from last semester. I am going to grade on a curve. I think Ed25519 is the most suitable, but if you have better suggestions, I am more than open for ideas. After all, I am a symmetric-key person.

itself by allowing AI reviewing, besides in special cases(!). Specifically, if I did not use AI in writing my paper, I should automatically be allowed to not have my paper reviewed by AI.

I don't see the univ. of Haifa trying to fight OpenClaude even if I can provide a proof a stealing. C) I remind you, these companies stole IP, broke the law, and essentially got away with it. Whatever the author does is his/her responsibility. The IACR should not expose >>>

release about "oh, this LLM stole something despite the Terms and Conditions". A) it is hard to prove the leak has happened (unless you are working for the LLM company, and even then). B) Any half decent LLM company has way more money then almost any university/medium company>>>

The author was hurt, and he can claim the IACR screwed him/her over. But the IACR has no contract with the AI company. The reviewer has some contract, usually through his company/university, but I am not sure many universities will either litigate or even hold a press >>>

but then you have a rebuttal/submit again). (b) even the question "are there any previous works doing X" reveals information about the contents of the paper. 2. If a reviewer used an LLM system in "private" mode, and the LLM leaked the contents, who is responsible? >>>

acknowledgement comparison). The reviewers part is quite horrible. The default is that one can ask the AI to review. 1. Obviously, there are things in which AI can help (e.g., verify proofs or look for relevant work), but (a) some time the AI is wrong (also reviewers, >>>

The @IACR_News has some new AI policy. There are mainly two parts - for authors and for reviewers. The authors part is quite good. Very clear distinction about what is the level of contribution that clarifies a lot of questions one may have (I really like the author vs. >>>

What a sucker was Morris from the first internet worm. He could have just said that his worm was written by AI, and he had no control over it...

We don't need a working AGI. We need a working PKI.

I am now an ML researcher. Thanks @AvitalShafran @eyalr0 @LukasFluri_ !

I wonder if there is a way to identify from the distance what is the brand of the biometric camera taking pictures at the airport. I mean, who identifies the identifier?!?!?

Reminder: On October 26th, we will commemorate the late Michael O. Rabin, Turing award laureate and one of the founding fathers of computer science in Israel in Tel Aviv University. Please help in letting your colleagues, especially Israeli ones, know about the event.

ישראלים אקדמאיים בחו"ל, זה בשבילכם: ב25-26 באוקטובר יתכנסו אלפי חוקרות, חוקרים ואנשי מקצוע מכל העולם ליומיים של מפגשים מדעיים, שיתופי פעולה, הזדמנויות מחקר חדשות וחשיבה פורצת דרך. אם עתיד המדע והמחקר בישראל חשוב לכם – הצטרפו אלינו! לפרטים נוספים והרשמה>>