Cynddl

12 ms·

1.
by Cynddl 12d ago
From the report: > A few hours after they find the site, [the agents] start probing it for cross-site scripting (XSS) vulnerabilities.
2.
by Cynddl 14d ago
> it seems fairly clear that A/I was providing services to organizations listed as Terrorist Organizations by the US, UK, and Canada Let's take the UK as example. The organisation mentioned, Palestine Action, was indeed banned
3.
by Cynddl 14d ago
> among the groups receiving those services was the PKK Do you have any evidence for that apart from a US executive order, the same that Banca Etica condemns for political motives?
4.
by Cynddl 1mo ago
> The providers did not acknowledge “any security implications arising from side channels or replay attacks.” All model providers acknowledged the receipt of our report and subsequently we were unable to launch the same attacks. I went s
5.
by Cynddl 2mo ago
I had to search these terms, does ADR mean ‘Architecture decision record’ and C4 the ‘C4 model’? How do you combine them in documentation?
6.
Thomson Reuters built its own AI model that now ranks among the best(thomsonreuters.com)
18 points by Cynddl 2mo ago|5 comments
7.
by Cynddl 2mo ago
Good question. Unfortunately, academic knowledge is widely ‘verboten’ already. Everything under paywall, researchers having to pay up to $10,000 to publish in open access in some venues, rare books unavailable even to top universities. Acce
8.
by Cynddl 2mo ago
The issue here is ACM focusing on licensing. A non profit would not be able to pay ACM for access. Hence why this policy is hypocritical: it gives more power to the larger players and undermines smaller actors in the field who have fewer re
9.
by Cynddl 2mo ago
This makes no sense whatsoever. How would a philosopher of science, or a social scientist who publish in the ACM apply for a patent? Or someone who builds software (software patent not so easy to get ;)). I have applied for patents before a
10.
by Cynddl 2mo ago
As a researcher with many articles in the ACM library, I have to say this is a masterclass in hypocrisy. Obviously, lawyers can decipher the terms of ACM publishing contracts and Creative Commons licences to determine if this will be accept
11.
by Cynddl 2mo ago
> For now these books are in corpuses of training data, but eventually I trust they will make their way to the rest of us. What makes you think they will? What would be the incentives for these companies to do so?
12.
by Cynddl 2mo ago
The 404media article mentions notably https://nltimes.nl/2026/06/25/rare-book-dealers-fear-tech-fi... which says: > The attachment contained 3,000 English-language titles organized by ISBN number, includin
13.
by Cynddl 2mo ago
Entropy is unfortunately a very bad metric to estimate if these identification techniques will scale. Plugging my work here as example: https://www.nature.com/articles/s41467-024-55296-6
14.
Hijacking Defensive Cyber AI Agents for Remote Code Execution(ainowinstitute.org)
2 points by Cynddl 2mo ago|0 comments
15.
by Cynddl 3mo ago
If you look at the bottom of the page, you’ll find guidelines that mention which content is welcomed: “Anything that good hackers would find interesting. That includes more than hacking and startups. If you had to reduce it to a sentence, t
16.
Our evaluation of OpenAI's GPT-5.5 cyber capabilities(aisi.gov.uk)
2 points by Cynddl 5mo ago|0 comments
17.
by Cynddl 5mo ago
Hi all, co-author here! Happy to answer any questions about our work.
18.
Making AI chatbots friendly leads to mistakes and support of conspiracy theories(theguardian.com)
93 points by Cynddl 5mo ago|80 comments
19.
by Cynddl 5mo ago
(Title edited, was slightly too long)
20.
by Cynddl 5mo ago
This sounds super interesting and relevant. I run a small cluster with H100s (often research projects with vLLM) and being able to see not just usage but efficiency would be great. I don't fully get the 100% utilisation vs. 1-10% real
21.
by Cynddl 5mo ago
It's not a zero-sum game, you can both protect people and reap the benefits of health data. Many countries have much safer approaches. UK Biobank typically leads with the scale of the data, but not with its infrastructure.
22.
by Cynddl 5mo ago
That's a very important point. The people who opt out first are typically not a random fraction of the population, and this makes it much harder to make any analyses with the resulting datasets: it gets very hard to know if your analys
23.
by Cynddl 5mo ago
Good catch! The data is everywhere, re-uploaded every week. I am aware of ~30 repositories that UK Biobank has asked GitHub to delete, and can still be found elsewhere online. They know the site, they have managed to delete data from that s
24.
by Cynddl 5mo ago
You mean giving anyone access to the data? Or open sourcing the code? If the latter, I think that's a generally a good practice. Security through obscurity is never good for public infrastructure. In this case, UK Biobank has now switc
25.
by Cynddl 5mo ago
They may have been leaked up to 197 times: https://biobank.rocher.lc/
26.
UK Biobank health data keeps ending up on GitHub(biobank.rocher.lc)
197 points by Cynddl 5mo ago|57 comments
27.
by Cynddl 5mo ago
yes, there’s an active area of research on web fingerprint, both attacks and defences. Look at conferences like PETS for instance
28.
by Cynddl 5mo ago
Is it me or they very carefully do not report performance on GPT-5.4 Pro, only the default GPT-5.4? They also very carefully left Anthropic models out of their comparison. I went back to the BixBench benchmark which they mentioned. I couldn
29.
by Cynddl 5mo ago
> Each case runs three agents: a Curator reads the advisory and builds an answer key, a Finder (the model under test) gets 24 shell steps to explore the code and write a structured report, and a Judge scores the blinded submission. The F
30.
by Cynddl 5mo ago
Once again an evaluation missing confidence intervals. “continued improvement” and “significant improvement” but without any significance testing is moot. With many colleagues (including from AISI themselves!), we recently reviewed 445 the

More ›