Ad/Marketing Communication
This legal article/report forms part of my ongoing legal commentary on the use of artificial intelligence within the justice system. It supports my work in teaching, lecturing, and writing about AI and the law and is published to promote my practice. Not legal advice. Not Direct/Public Access. All instructions via clerks at Doughty Street Chambers. This legal article concerns AI Law.
Introduction
I apologise in advance, this post is likely to be longer than most. It is not every day that the European Court of Human Rights consider artificial intelligence and especially the potential for AI Rights.
The case of Ships Waste Oil Collector B.V. and Others v. the Netherlands (applications nos. 2799/16, 2800/16, 3124/16 and 3205/16) concerned the transmission of “by-catch” data, lawfully obtained in criminal investigations through telephone tapping, to another law enforcement authority, the Competition Authority, that had used those data in unrelated administrative investigations into the applicant companies’ involvement in price-fixing. The relevant documents can be read here. I would suggest everyone interested in this topic read the case in full.
Brief Summary
Several Dutch companies were involved, each facing hefty fines for alleged price-fixing under the Competition Act. The dispute stemmed from the fact that telephone data acquired under criminal procedure, aimed at investigating different suspected offences, were forwarded to the Netherlands Competition Authority (NMA). The applicant companies argued that this ‘by-catch’ data should never have been handed over or used for a non-criminal purpose, such as competition enforcement.
The relevant humans rights that were engaged were:
- Article 8 (right to respect for correspondence)
- Article 13 (right to an effective remedy).
The Judges
A Grand Chamber hearing was held on 6 March 2024. Judgment was given by the Grand Chamber of 17 judges, composed as follows:
Marko Bošnjak (Slovenia), President,
Arnfinn Bårdsen (Norway),
Lado Chanturia (Georgia),
Mattias Guyomar (France),
Georges Ravarani (Luxembourg),
Carlo Ranzoni (Liechtenstein),
Georgios A. Serghides (Cyprus),
Tim Eicke (the United Kingdom),
Lətif Hüseynov (Azerbaijan),
Jovan Ilievski (North Macedonia),
Jolien Schukking (the Netherlands),
Raffaele Sabato (Italy),
Saadet Yüksel (Türkiye),
Lorraine Schembri Orland (Malta),
Kateřina Šimáčková (the Czech Republic),
Davor Derenčinović (Croatia),
Oddný Mjöll Arnardóttir (Iceland),
and also Johan Callewaert, Deputy Grand Chamber Registrar.
The Decision
The Court recognised that sharing criminally obtained data with non-criminal authorities can be highly intrusive and requires a clear, foreseeable legal basis. However:
BY 12 votes to 5, no violation of Article 8 in respect of Ships Waste Oil Collector B.V., Burando Holding B.V. and Port Invest B.V.;
BY 10 votes to 7, no violation of Article 8 in respect of Janssen de Jong Groep B.V., Janssen de Jong Infra B.V. and Janssen de Jong Infrastructuur Nederland B.V.;
BY 15 votes to 2, no violation of Article 13
So ultimately, the judges concluded that while the phone tapping itself was lawful, the subsequent data transmission to the NMA was also permissible under Dutch law, even if some steps (like prior ‘exploratory’ data access) proved controversial. A majority ruled that the statutory safeguards, along with post-transmission judicial review, were enough to meet the Convention’s standards.
However, what I found particularly interesting was the dissenting opinion of Judge Serghides where he observed some important AI related issues in paragraphs 47 to 49.
Comment
I’ve been reading Judge Serghides’ comments very closely. These insights stood out to me because they’re exactly the kind of thoughtful discussions I often have with colleagues, but to see them presented so clearly in an actual judgment was particularly striking.
In this blog post, I’m trying to explain some complex legal issues and terms in a straightforward way, making sure anyone can follow along, even without a legal background. But I realise this approach risks oversimplifying or skipping important details. So, before getting into the main discussion, I’ll briefly outline a few key points from the comments. Hopefully, this will make the critical ethical, philosophical, and legal issues around AI and human rights easier for everyone to understand.
The Judge states at paragraph 47:
“Not only does the French term “personne physique” bar companies from having victim status through the “person” window, but it may also bar the Court from affording victim status to different forms of artificial intelligence through this window in its future case-law.”
Let’s unpack what that means.
- What is Personne physique”? In French, this phrase means “natural person.” Under the European Convention on Human Rights, the French text specifically uses “personne physique” when referring to the individual who can complain of a violation of their rights. Essentially, it’s talking about human beings, not companies or other entities.
- What is the relevance of “companies”? The paragraph underscores that since the legal phrase “personne physique” means we are talking about human beings, companies are already barred from claiming “person” status as if they were a human. By extension, “it may also” mean any non-human entity like different forms of artificial intelligence.
- “Victim Status”? Going forward, if an AI system or “robot” attempts to go to the Court (the European Court of Human Rights) claiming it is a victim of a rights violation, it seems the current language, “personne physique”, may stop them at the door. This is because that term is specifically about human persons, not AI, not robots, not corporate structures.
- The “person” window? This phrase is used figuratively. Think of Article 34 of the European Convention on Human Rights as having certain “windows” or “gateways” through which an applicant can enter. One of these is the “person (human) window”, but that is limited to real human beings. If an AI tries to use that path, it’s likely blocked because the law’s wording was written for humans only.
At paragraph 48, the Judge continues.
“If forms of artificial intelligence possessing some of the functions of natural persons, like robots, seek protection before the Court, the term “personne physique” will prevent them from passing through the “person” window.”
By “Some of the functions of natural persons,” we’re talking about hypothetical AI that can do a good chunk of what humans do. For example, thinking, communicating, reasoning, and maybe making decisions that appear human-like. Yet no matter how advanced these systems become, they’re not biological humans. At least under the current language, they’d likely be barred.
Now paragraph 49 is where it gets particularly thought-provoking:
“If, one day, member States decide to provide some kind of protection for artificial intelligence, it would be better done through a new Protocol. A new Protocol would also offer the opportunity to firmly and clearly set out a legal basis for the claims of entities other than natural persons, including companies. However, one can only wonder how such developments in the future, if they came to be, could be reconciled with the Convention’s primary aim for the effective protection of human rights and the principle of human dignity! If, in the future, artificial intelligence has rights similar or adjacent to human rights safeguarded by the Convention, as granted by a new Protocol to the Convention, how could one be certain that it would itself respect human rights, the rule of law and the principle of democracy. Though it may be conceivable that autonomous artificial intelligence may in the future benefit from the protection of rights similar or adjacent to human rights contained in the Convention, at the same time, it is crucial that this goes hand in hand with the capacity and obligation of such artificial intelligence to respect human rights, the rule of law and the principle of democracy. However, I cannot continue further this interesting, though theoretical and philosophical discussion, since this is not the issue here.”
The Judge ends with that final line, indicating these words are more theoretical and not a binding precedent. I’d add, however, that it feels like the Judge is nudging us to think about what it all means. And I agree, this discussion is extremely important.
Essentially, the Judge suggests that if we ever want to give AI or any non-human entity formal standing before the Court, we’d have to create a new Protocol. In simple terms, a Protocol is an amendment or an added-on treaty to the main Convention, clarifying or expanding rights in some way.
But how would that fit with the Convention’s “primary aim …the effective protection of human rights” and the principle of “human dignity!”?
Pausing there, in my experience it is rare to see an exclamation mark used in a judgment like this. I don’t think that exclamation mark is random. It drives home the gravity of these questions.
Finally, the big question: if AI is accorded rights adjacent to human rights, how can we be certain it will also respect human rights, the rule of law, and democracy? That question resonates with me. It’s central, and I can’t fully address it in this one blog post. I’ll write another more extensive piece at some point soon.
However, for immediate context on potential dangers linked to this topic, readers might find my other blog posts insightful which cover AI hallucinations, AI Parasites and Cognitive Security and Secret Deception. I’m also finalising some writing on discrimination and AI bias. Unless these dangers are smoothed out, we’ll be treading dangerously.
My view is that these concerns need addressing, and, with enough time, I’m confident the common law, open dialogue, and regulation can provide solutions.
However, I fear, we don’t have that time.
Before deep learning took off around 2010, the amount of computation used to train AI systems doubled about every 21 months. But, this has accelerated significantly with the rise of deep learning, now doubling roughly every six months.
If you are interested in civil liberties and human rights, we need more candid conversations about where this could be heading. Please reach out, either through this blog or LinkedIn, if these issues spark your interest. I welcome all perspectives, whether you share my concerns or believe it’s safe (perhaps even commendable) to extend human rights to AI.
We’re all trying to understand where this is going, and collectively we have a much better chance of steering it toward the right outcome.
