Marcel Keller's Avatar

Marcel Keller

@mkskeller.bsky.social

Cryptographer who likes to implement multi-party computation and works for CSIRO's Data61. Views my own. πŸ³οΈβ€πŸŒˆπŸ‡¨πŸ‡­πŸ‡¦πŸ‡Ί (he/him)

476 Followers  |  302 Following  |  42 Posts  |  Joined: 05.02.2024  |  1.828

Latest posts by mkskeller.bsky.social on Bluesky

Call for submissions: #TPMPC2026 (Theory & Practice of MPC)

Submit your latest and coolest results by March 2, 2026.

Aarhus, Denmark, May 18–22, 2026.

Monday: MPC security in practice.

Friday: Symposium celebrating Ivan DamgΓ₯rd’s work.

Links in comments.

10.02.2026 08:18 β€” πŸ‘ 5    πŸ” 5    πŸ’¬ 1    πŸ“Œ 0
Cartoon showing a street in the rain. A man on a bike is delivering food, while another courier is delivering packages from Amazon. Two other workers are collecting garbage. Inside one of the houses on the street, we see robots labeled 'AI' sitting dry and warm, engaged in making a paining, playing the violin and writing.

Cartoon showing a street in the rain. A man on a bike is delivering food, while another courier is delivering packages from Amazon. Two other workers are collecting garbage. Inside one of the houses on the street, we see robots labeled 'AI' sitting dry and warm, engaged in making a paining, playing the violin and writing.

The robot apocalypse hasn't happened yet, but still I can't escape the feeling that something has gone horribly wrong... Cartoon for Dutch newspaper @trouw.nl.

More of my work for Trouw: www.trouw.nl/cartoons/tje...

#ArtificialIntelligence #creativity #work #GenerativeAI

02.02.2026 07:38 β€” πŸ‘ 515    πŸ” 226    πŸ’¬ 9    πŸ“Œ 12
Person: say, i am alive.

Computer: I am alive.

Person: oh my god.

Person: say, i am alive. Computer: I am alive. Person: oh my god.

01.02.2026 23:48 β€” πŸ‘ 23376    πŸ” 5153    πŸ’¬ 84    πŸ“Œ 140

The number of people arguing, following the #NeurIPS2025 case, that a bit of systematically bad scholarship and borderline academic misconduct is OK because doing things the right way is tedious and time-consuming is concerning.

Yes, doing things well might be boring. But doing them badly is... bad

24.01.2026 22:25 β€” πŸ‘ 56    πŸ” 8    πŸ’¬ 1    πŸ“Œ 0
Preview
I’m going to halve my publication output. You should consider slow science, too If we don’t slow down, the research enterprise is going to crash, argues Adrian Barnett.

Advice for Jens Eisert.
doi.org/10.1038/d415...

20.01.2026 07:06 β€” πŸ‘ 12    πŸ” 4    πŸ’¬ 1    πŸ“Œ 0
Preview
Petition Restore Fully Free and Open Access to the ACM Digital Library Restore Fully Free and Open Access to the ACM Digital Library

Consider signing this petition regarding restoring free access to the ACM digital library: www.ipetitions.com/petition/res...

09.01.2026 08:08 β€” πŸ‘ 8    πŸ” 7    πŸ’¬ 0    πŸ“Œ 0

Deadline is January 7. Still a few days left to apply...

02.01.2026 10:17 β€” πŸ‘ 2    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0
Preview
β€˜All brakes are off’: Russia’s attempt to rein in illicit market for leaked data backfires Russian state has tolerated parallel probiv market for its convenience but now Ukrainian spies are exploiting it

Privacy issues are unlikely to go away in 2026. www.theguardian.com/world/2025/d...

26.12.2025 12:28 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Release Expected communication cost in compiler Β· data61/MP-SPDZ Expected communication cost in compiler Semi-honest option of Rep4 Reduced communication for preprocessing in Dealer protocol Option of choosing SoftSpoken parameter at run-time BERT functionality ...

The latest version of MP-SPDZ can predict the communication cost when compiling high-level code. github.com/data61/MP-SP...

24.12.2025 04:21 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Decentralized Key Generation for Zama Mainnet Has Successfully Completed The Decentralized Key Generation (DKG) ceremony for the Zama mainnet has completed successfully.

Fully robust MPC protocol over Galois Rings, executing a TFHE DKG protocol amongst 13 parties, taking 55 hours, with 10.5 Terabytes of data sent per party....

www.zama.org/post/decentr...

19.12.2025 11:36 β€” πŸ‘ 11    πŸ” 2    πŸ’¬ 1    πŸ“Œ 0

Come work with Peter Scholl @schollster.bsky.social and me in Aarhus!

17.12.2025 12:17 β€” πŸ‘ 4    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0
Preview
Cryptographers Show That AI Protections Will Always Have Holes | Quanta Magazine Large language models such as ChatGPT come with filters to keep certain info from getting out. A new mathematical argument shows that systems like this can never be completely safe.

Cryptography shows that AI filters will never work....

www.quantamagazine.org/cryptographe...

Nice application of crypto to AI, and shows that AI has inherent limitations.

16.12.2025 07:08 β€” πŸ‘ 3    πŸ” 3    πŸ’¬ 1    πŸ“Œ 0
Preview
Cryptographers Show That AI Protections Will Always Have Holes | Quanta Magazine Large language models such as ChatGPT come with filters to keep certain info from getting out. A new mathematical argument shows that systems like this can never be completely safe.

If you swap each letter in β€œbomb” with the next letter in the alphabet, you’ll get β€œcpnc.” Recently, scientists showed that and other methods can bypass filters on LLMs like Gemini, DeepSeek and Grok. @peterha2l.bsky.social reports: www.quantamagazine.org/cryptographe...

10.12.2025 15:42 β€” πŸ‘ 23    πŸ” 8    πŸ’¬ 0    πŸ“Œ 1

On the internet, age verification doesn't know you're a dog.

10.12.2025 04:34 β€” πŸ‘ 567    πŸ” 114    πŸ’¬ 15    πŸ“Œ 2

"Maybe the name EasyCrypt isn't quite deserved." - @cryptojedi.bsky.social at his Asiacrypt keynote

10.12.2025 03:26 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
MPC in the Wild is a research project that aims to study the real-world implementation security of MPC protocols, in particular, the gap between assumptions on the theory side and implementation choices made in practice.

Interested in the security of MPC implementations?

Peter Scholl (@schollster.bsky.social) and I are looking for research interns to study the theory/practice gap for MPC. We can host multiple interns in Aarhus for the summer 2026.

More information: mpcinthewild.github.io

Help us spread the word!

06.12.2025 16:21 β€” πŸ‘ 20    πŸ” 10    πŸ’¬ 0    πŸ“Œ 1
Preview
Major AI conference flooded with peer reviews written fully by AI Controversy has erupted after 21% of manuscript reviews for an international AI conference were found to be generated by artificial intelligence.

www.nature.com/articles/d41...

03.12.2025 08:13 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Stop Hacklore!

πŸ“’ Announcing hacklore.org πŸ“’

It’s time to retire outdated cyber advice! More than 80 cybersecurity veterans have signed an open letter urging a shift from folklore to guidance that actually helps people avoid the most common attacks. πŸ”

Blog: medium.com/@boblord/let...

Site: www.hacklore.org

24.11.2025 15:05 β€” πŸ‘ 107    πŸ” 64    πŸ’¬ 6    πŸ“Œ 13
OpenAlex

At least they make it easy to find out why. It seems they confuse you with Moni Naor on this CiC paper: openalex.org/works/W44092...

21.11.2025 04:00 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

One of the motivations for MP-SPDZ was not getting bogged down by this kind of arguments when they are just hypothetical. Every use case has its own trade-offs as the IACR just found out.

21.11.2025 03:48 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Just saw an extended version

20.11.2025 23:25 β€” πŸ‘ 197    πŸ” 79    πŸ’¬ 3    πŸ“Œ 11
Preview
Massive Leak Shows Erotic Chatbot Users Turned Women’s Yearbook Pictures Into AI Porn Chatbot roleplay and image generator platform SecretDesires.ai left cloud storage containers of nearly two million of images and videos exposed, including photos and full names of women from social me...

This is a nuts story. An AI chatbot and image platform left millions of images exposed. They show what people are actually using the AI for: taking random women's yearbook, graduation, and social media photos and making super realistic hardcore porn with them
www.404media.co/ai-porn-secr...

19.11.2025 15:28 β€” πŸ‘ 2127    πŸ” 829    πŸ’¬ 96    πŸ“Œ 326
Preview
DOJ Issued Seizure Warrant to Starlink Over Satellite Internet Systems Used at Scam Compound A new US law enforcement initiative is aimed at crypto fraudsters targeting Americansβ€”and now seeks to seize infrastructure it claims is crucial to notorious scam compounds.

NEW: The US Department of Justice issued a warrant demanding SpaceX seize and disable Starlink devices and accounts being used at a scam compound in Myanmar.

At least 9 Starlink devices were linked to a crypto scam that stole $6 million people, an FBI affidavit says

Story with @lhn.bsky.social

14.11.2025 20:55 β€” πŸ‘ 244    πŸ” 94    πŸ’¬ 16    πŸ“Œ 13

A staggering statistic: "North American researchers were charged over US$2.27 billion by just two for-profit publishers. The Canadian research councils and the US National Science Foundation were allocated US$9.3 billion in that year." What are we doing?

12.11.2025 13:58 β€” πŸ‘ 107    πŸ” 59    πŸ’¬ 3    πŸ“Œ 7
Preview
Add BERT to ML Library by hiddely Β· Pull Request #1736 Β· data61/MP-SPDZ This is an implementation of Huggingface's BERT transformer that I've been meaning to contribute back to MP-SPDZ. I implemented it as part of the evaluation of this work. The implementation...

Thanks to Hidde Lycklama, MP-SPDZ now supports BERT inference. github.com/data61/MP-SP...

07.11.2025 04:10 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Slide is titled: You don't need to use LLMs. 

Science is a process of collaborative meaning making, by which we try to understand the world

Even if AI were perfect, we rely on it at our peril β€” it is not science if we (i.e., humanity as a whole) do not understand and cannot recapitulate all parts of it

I very, very rarely use LLMs myself. You can give yourself permission not to. Don’t FOMO yourself into it

Slide is titled: You don't need to use LLMs. Science is a process of collaborative meaning making, by which we try to understand the world Even if AI were perfect, we rely on it at our peril β€” it is not science if we (i.e., humanity as a whole) do not understand and cannot recapitulate all parts of it I very, very rarely use LLMs myself. You can give yourself permission not to. Don’t FOMO yourself into it

Conclusion: Don't rely on something you don't understand and can't control

If you must use LLMS:

1. Treat them like you would an intern: only use them for things you can easily and thoroughly check

2. Make your process as robust as possible

3. Be aware of your own (human) cognitive biases

Conclusion: Don't rely on something you don't understand and can't control If you must use LLMS: 1. Treat them like you would an intern: only use them for things you can easily and thoroughly check 2. Make your process as robust as possible 3. Be aware of your own (human) cognitive biases

Getting nervous for the talk I'm about to give at a workshop about "using AI to drive impact" which features slides such as these.

06.11.2025 20:41 β€” πŸ‘ 379    πŸ” 90    πŸ’¬ 26    πŸ“Œ 11
Preview
Homes could be $100k cheaper in 10 years with one change to zoning rules, Australian report finds Allowing three-storey townhouses and apartments to be built across all residential land would unlock more than 1m homes in Sydney alone, according to modelling

Why @sydney.yimby.au: www.theguardian.com/australia-ne...

05.11.2025 10:55 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Remarkable, but important to notice that the change only applies to review/survey and position papers

04.11.2025 02:10 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
The AWS Outage Bricked People’s $2,700 Smartbeds When Amazon Web Services went offline, people lost control of their cloud-connected smart beds, getting stuck in reclined positions or roasting with the heat turned all the way up.

The AWS outage bricked people's $2,700 smartbeds, leaving them in a reclining position or with the heat up. People were unable to use them, basically. A GitHub repo exists that lets people operate their beds without reliance on the normal infrastructure

www.404media.co/the-aws-outa...

22.10.2025 13:43 β€” πŸ‘ 634    πŸ” 195    πŸ’¬ 72    πŸ“Œ 358

Our protocol is simple & modular: it works in the "Arithmetic Black Box" model, and inherits the security properties of the ABB instantiation (passive/active, static/adaptive, etc.).

Using SPDβ„€β‚‚π‘˜ for ABB, we get about 20,000x and 37x better throughput and latency than SoTA for dishonest majority.

14.10.2025 15:40 β€” πŸ‘ 2    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

@mkskeller is following 20 prominent accounts