Due to acting like an irrational gambling machine, I agree it can have unwanted indirect discrimination effect in general. But it will probably not differentiate "on the grounds of religion or belief, disability, age or sexual orientation". It is possible, but that would take a lot of work for the lawyers to prove to the court.

I believe the more interesting part is that the EU AI Act (still not in force in this regard until 2 December 2027). This will be clearly a high-risk AI system: "AI systems intended to be used for the recruitment or selection of natural persons, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates".

Which does not mean prohibited, but it could later turn out that LLMs will be excluded from being used in high-risk AI use cases (falling under article 6 with no exemptions).

Considering that none of the standards are published yet, I have absolultely no idea how they will ensure compliance with the following parts of Article 10 when using LLMs for such tasks: "(f) examination in view of possible biases that are likely to affect the health and safety of persons, have a negative impact on fundamental rights or lead to discrimination prohibited under Union law, especially where data outputs influence inputs for future operations; (g) appropriate measures to detect, prevent and mitigate possible biases identified according to point (f)"

I don't think that's technically possible to do so with LLMs in general at the moment, even with the full cooperation of the model providers. Maybe you can do some meaningful audits for smaller models. But the EU AI Act may end up excluding all the generic "using-LLM-but-not-entirely-sure-why" vibe coded approaches from high-risk use cases (in Annex III). Which would make sense.

https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng

And for non-common law jurisdictions, even the UK level of publicity of judgments is rarely available - anonymisation is a prerequisite: https://homoki.net/en/2024/02/01/On_Anonymisation_of_court_d...

So, for many countries, it's not just that sealed documents are not accessible, many people in traditional democracies couldn't imagine and bear this level of publicity for every court document like you have in the US.

In Germany, you also have the Oberlandesgericht and the Bundesgerichtshof as a higher venue for civil cases (and the Constitutional Court as well). So, this is not like "the law has changed" at all. (I'm not a German qualified lawyer, but qualified in a close enough continental system, and on my way to be qualified in a common law jurisdiction soon, that's why I dare to comment on this aspect)

It is addressed to lawyers and bars/law societies (which, in most countries, also regulate lawyers).

It is a clear and deliberate step to educate legal professionals on more technical matters that were previously rarely a focus of their training.

The purpose of the paper is to better acquaint lawyers with their options, including the possibility of using local models and understanding the trade-offs.

CCBE was created by the bars/law societies as a representative body of the profession in Brussels, to whom the EU institutions can turn.

I believe this is an important step in toward better dissemination of the use of AI models. What makes it notable is the level of technical detail that will hopefully help lawyers formulate what they need - comparing the options of on-premises, colocation, IaaS or fully managed/API), and giving indicative hardware budgets (dated September 2025) ranging from ~€2,000 to €20,000 for local inference boxes. It also discusses quantisation (FP16 vs INT8 vs INT4), context-length trade-offs, and tokens-per-second thresholds for interactive use.

Curious to hear your views on whether there is any point in going in this direction. Or if you believe anything important is missing or incorrect.

Comments URL: https://news.ycombinator.com/item?id=47792486

Points: 3

# Comments: 1

I don't think people are afraid of doctors using imperfect tools. That is the easier part. But that will not solve the problem of too many patients for a single doctor and what leads to the lack of empathy. This was a problem even before AI. It seems society does not have empathy for these kind of "professional problems". Offering tools instead of humans is an even riskier approach, not for that particular individual, but for how society tends to build trust and empathy. We tend to see everything now as a problem with a technical solution because we only have confidence in solving technical problems.