AI in business: the options, the dangers and how to get it right

Three ways to use AI, and one question that decides between them

Most businesses meet AI the same way. Someone signs up for a chatbot, it writes a decent email in four seconds, and within a month half the office is using it for everything. Nobody decided anything. It just happened.

That is the moment to step back, because there are really only three ways to put AI to work, and each one answers a single question differently: where does your data go when somebody types into the box?

• Cloud AI. You use a provider's model over the internet. Your prompt leaves your building, is processed on their servers and comes back as an answer.
• Self hosted AI. You run an open weight model on hardware you control. The prompt never leaves your network.
• Hybrid AI. A routing layer sends each request to one or the other, depending on what is in it.

This guide walks through all three, what goes wrong when people get careless, and what the law says in Ireland and across the EU.

First, a reality check from the Central Statistics Office (CSO), whose survey of enterprises with 10 or more persons engaged asks every year about AI. The share using at least one AI technology rose from 8% in 2023 to 15% in 2024 and 20% in 2025.¹² Use grows with size: in 2025, 17% of small enterprises (10 to 49 persons engaged) used AI, against 29% of medium enterprises (50 to 249) and 58% of large ones.²

The vertical axis is the share of enterprises using at least one AI technology, the horizontal axis the size band and year. Use has more than doubled in two years, yet four in five enterprises with 10 or more staff still report none. You are not late. You are early enough to do it properly.

Throughout, I use one invented business. Ballybricken Bookkeeping is a 14 person bookkeeping practice in Waterford, run by Eimear, at ballybricken.example. It does payroll, VAT and year end accounts for about 200 small clients. Nobody real, every number made up.

Cloud AI: the most capable option with the least setup

These are the names you have heard: ChatGPT from OpenAI, Claude from Anthropic, Gemini from Google and Copilot from Microsoft. Your request is processed in the provider's cloud and the answer comes back in seconds. No hardware, nothing to install, no model to update. You pay a subscription or per token (a token is a chunk of text, roughly a short word).

They draft copy, summarise reports, analyse documents, translate, and write and explain code. The providers update them constantly, which is why I do not list what any particular model can do: claims like that age in weeks. Check the provider's current documentation for the model you actually pay for.

The strongest argument is hardware. A large model needs serious memory on a graphics card (GPU) just to load, and cloud providers spread that cost across millions of users.

Worked example: what a cloud seat really costs Ballybricken

Eimear wants all 14 staff on a business plan of a cloud chatbot. Assume €25 per seat per month (the example's assumption, not any provider's price list).

$$14 \times 25 \times 12 = 4{,}200$$

Here $14$ is the number of seats, $25$ the assumed price per seat per month in euro, $12$ the months in a year, and $4{,}200$ the yearly cost in euro.

For €4,200 a year Ballybricken gets the most capable models with no maintenance. The rest of this guide is about what that does not cover.

Self hosted AI: open weight models on your own hardware

The alternative is to run a model on infrastructure you control. "Open weight" means the trained model itself (its weights, the billions of numbers that make it work) is published for anyone to download and run. Some well known families, each with its own licence:

• Llama from Meta. Commercial use is allowed under Meta's licence, with a condition: a business with more than 700 million monthly active users must ask Meta for a separate licence.³
• Mistral. A French company. When it released Mistral 7B in September 2023, it reported that the 7.3 billion parameter model beat Meta's Llama 2 13B, nearly twice its size, on all the benchmarks it tested, and published it under the Apache 2.0 licence.⁴ Size is not everything.
• DeepSeek. The DeepSeek R1 reasoning model is published under the MIT licence.⁵
• Qwen from Alibaba, Gemma from Google and Falcon from the Technology Innovation Institute in Abu Dhabi round out the familiar names.

Which one is good enough for your work is not something a list can tell you. Test two or three candidates on 20 real, anonymised tasks from your own business and score the answers yourself.

The tools developers use

If you build AI into your own software, these tools come up again and again:

• Ollama downloads and runs a model with a single command, the easiest way to try local AI.
• llama.cpp runs models directly on your hardware with fine control over every setting, and vLLM serves models to many users at once.
• Hugging Face Transformers is the standard Python library for working with pretrained models.
• LangChain chains prompts and tools into multi step workflows, and LlamaIndex connects a model to your own documents and databases, the pattern called retrieval augmented generation (RAG).

Worked example: how much memory a model needs

Memory is roughly parameters times bytes per parameter. Hugging Face's documentation: models usually load at 2 bytes per parameter, and quantisation (storing each weight in fewer bits) cuts that, so an 8 billion parameter model needs about 16GB at full size, 8GB at 8 bits and 4 to 5GB at 4 bits, at some cost in accuracy.⁶

Eimear's developer looks at a 70 billion parameter model, quantised to 4 bits. Four bits is half a byte:

$$70{,}000{,}000{,}000 \times \frac{4}{8} = 35{,}000{,}000{,}000 \text{ bytes}$$

Here $70{,}000{,}000{,}000$ is the number of parameters, $\frac{4}{8}$ converts 4 bits into half a byte per parameter, and $35{,}000{,}000{,}000$ bytes is 35GB for the weights alone. The model also needs working memory for the conversation itself, so in practice it wants around 40GB of GPU memory. That is professional hardware or several cards working together.

A 7 billion parameter model at 4 bits needs $7{,}000{,}000{,}000 \times \frac{4}{8} = 3.5$ billion bytes, 3.5GB, which fits on a consumer graphics card with 8GB. That is the trade: smaller models run on modest hardware, larger ones handle harder reasoning.

The privacy problem: when convenience becomes a liability

Every prompt sent to a cloud AI leaves your infrastructure. It is processed on servers you do not control and, depending on the plan and the settings, may be stored or used to train future models.

That last point depends on the contract, and it differs between consumer and business plans. Anthropic, for example, changed its consumer terms in August 2025 to let users of its Free, Pro and Max plans choose whether their chats are used to improve its models, and said the change did not apply to its commercial terms or to API use.⁷ The person in accounts who signed up with a personal email is on the consumer side of that line.

In May 2023 Samsung restricted the use of generative AI tools such as ChatGPT on company computers after staff uploaded sensitive code to ChatGPT, and said it was looking for ways to use generative AI safely.⁸

In August 2024 the Dutch data protection authority, the Autoriteit Persoonsgegevens (AP), warned employers after receiving breach reports caused by AI chatbots: an employee of a GP practice had entered patients' medical data into a chatbot, and an employee of a telecoms company had entered a file that included customer addresses.⁹ By December 2025 the AP said it had received dozens of such reports across 2024 and 2025, with more in 2025 than the year before.¹⁰

The case that prompted that warning was the Dutch city of Eindhoven. A check covering 30 days, from 23 September to 23 October 2025, found staff had uploaded files containing personal data to public AI websites: documents under the Youth Act, internal reports and job applicants' CVs. The city could not say how many files were affected in total, because the AI sites kept uploads for only 30 days. It blocked public AI websites, moved staff to a tool inside its own environment and asked OpenAI to delete the files.¹¹

What careless use looks like in practice

Nobody means any harm.

Finance. Someone asks a chatbot to tidy a profit and loss statement or draft investor notes and pastes in the real figures. That transfer may break internal policy, client contracts or market rules on inside information.

Data and client confidential information. A data scientist pastes a sample of a client's dataset, with names, emails and purchase histories, into a chatbot. Personal data has just gone to a third party without a proper processing agreement, without telling the people it is about and very likely without a lawful basis under Article 6 of the GDPR.¹²

Developers and keys. A developer pastes code into an AI assistant for help with a bug, with API keys, database passwords or payment gateway credentials buried in it. Those secrets have been transmitted and possibly logged, and a leaked key works for anyone who holds it. Who answers for code an assistant helped write is a question of its own, and I take it apart in AI writes the code. Who is responsible for it?, and what these assistants can and cannot do for a developer in Will AI like ChatGPT ever replace developers?

I've seen this kind of carelessness first hand. In my work with clients I came across a developer who had uploaded over 14,000 customer records to GitHub, where they stayed publicly accessible until I found them. The same habit that leads to mistakes in a code repository happens with AI tools, often with even less awareness of the risk.

Worked example: what one careless paste puts at risk at Ballybricken

A junior bookkeeper at Ballybricken is reconciling a client's payroll. The spreadsheet will not balance, so she pastes 40 rows into a free chatbot she signed up for with her personal email, and asks what is wrong.

Count what just left the building:

1. 40 employees' names, so 40 people's personal data.
2. Their salaries and PPS numbers (Personal Public Service Numbers), which identify them and could enable fraud.
3. The client's IBAN from the header row.
4. The client's name, which Ballybricken's engagement letter promises to keep confidential.

Now the obligations: no processing agreement with the provider, 40 employees never told, a client contract broken, and Eimear has to decide whether to report a breach to the Data Protection Commission. Nobody had told the bookkeeper where the line was.

Putting a number on the risk with expected value

"It might go wrong" is hard to budget for. Expected value turns a risk into a figure you can weigh against the cost of preventing it. It is the sum of every outcome times its probability:

$$E[C] = \sum_{i} p_i \, c_i$$

Here $E[C]$ is the expected cost per year, $p_i$ the probability of outcome $i$ in a year, $c_i$ the cost of that outcome in euro, and $\sum_{i}$ means add up across all outcomes. The idea is set out from scratch in Foundations: expected value.

Worked example: is a policy and a routing layer worth it for Ballybricken?

The figures are the example's assumptions, not measured rates. Eimear assumes that without any controls there is a 10% chance in a year of an incident like the payroll paste that becomes serious, and that a serious incident costs €30,000 (time spent, legal advice, notifying people and one lost client). There is a 90% chance of no serious incident, which costs nothing.

$$0.10 \times 30{,}000 + 0.90 \times 0 = 3{,}000$$

Here $0.10$ is the assumed probability of a serious incident, $30{,}000$ its assumed cost in euro, $0.90$ the probability of no serious incident, $0$ its cost, and $3{,}000$ the expected cost per year in euro.

With a written policy, training and a routing layer that catches the obvious personal data, she assumes the probability drops to 2%:

$$0.02 \times 30{,}000 = 600$$

Here $0.02$ is the assumed probability with controls and $600$ the expected cost per year in euro. The controls are worth up to $3{,}000 - 600 = 2{,}400$ euro a year in expected terms.

Expected value averages over many years. If losing the one big client would sink the practice, the average does not help in the year it happens. That is an argument for the controls, not against them.

The legal reality: GDPR, the AI Act and the Irish regulator

When staff enter anything that identifies a person into a cloud AI tool, that is processing of personal data, and the GDPR applies. It is an EU regulation, so it applies in Ireland directly, as it does in every member state.

Three articles do most of the work. Article 6 requires a lawful basis for any processing. Article 28 requires a contract with any processor handling personal data on your behalf, with set terms. Article 44 allows transfers of personal data to a country outside the European Economic Area only under the conditions in that chapter of the regulation.¹²

The transfer point is less dramatic than it was. In July 2023 the European Commission decided that the EU US Data Privacy Framework gives adequate protection for personal data sent to US organisations certified under it.¹³ In September 2025 the EU General Court dismissed a challenge to that decision.¹⁴ A free account opened with a personal email still comes with no processing agreement at all.

The fines, worked through for a small firm

Under the EU GDPR the most serious breaches, including those of the basic principles and the transfer rules, carry fines of up to €20 million or 4% of worldwide annual turnover, whichever is higher, under Article 83(5).¹² That is the regime that applies in Ireland.

Ballybricken's turnover is €1.2 million (an invented figure):

$$0.04 \times 1{,}200{,}000 = 48{,}000$$

Here $0.04$ is the 4% turnover cap, $1{,}200{,}000$ Ballybricken's annual turnover in euro, and $48{,}000$ the result in euro. Because the rule says "whichever is higher", the ceiling for Ballybricken is the fixed €20 million, not €48,000. The 4% figure only becomes the larger one above a turnover of $20{,}000{,}000 \div 0.04 = 500{,}000{,}000$ euro. Nobody expects a bookkeeper to be fined €20 million, but the ceiling shows how seriously the law takes it.

Fines are also less settled than headlines suggest. In December 2024 Italy's data protection authority, the Garante, fined OpenAI €15 million over ChatGPT, citing among other things the training of the model on personal data without an adequate legal basis. A court in Rome upheld OpenAI's appeal against that decision in a ruling published on 18 March 2026.¹⁵ The law is still being tested, case by case.

The EU AI Act, with its real dates

The EU AI Act is Regulation (EU) 2024/1689. It entered into force on 1 August 2024, the twentieth day after its publication, and applies in stages: the general date is 2 August 2026, the prohibited practices and the AI literacy duty applied from 2 February 2025, and the rules for general purpose AI models from 2 August 2025.¹⁶

Processing sensitive data does not make a system high risk. High risk is defined by what a system is used for, listed in Annex III: among others biometrics, critical infrastructure, education, employment (for example tools that filter job applications or evaluate candidates), access to essential services, and law enforcement.¹⁶ A chatbot that drafts emails is not high risk because someone pastes personal data into it. That is a GDPR problem.

The dates for high risk systems have since moved. The Digital Omnibus on AI, Regulation (EU) 2026/1744, published on 24 July 2026, delays the high risk obligations to 2 December 2027 for the systems listed in Annex III and to 2 August 2028 for AI built into products already covered by EU safety law. It also softened the AI literacy duty: providers and deployers must now "take measures to support the development of AI literacy of their staff", and the duty "does not require providers or deployers to guarantee any specific level of AI literacy of any individual".¹⁷ I cover what this means for software builders in the EU AI Act for SaaS builders, and the German and Austrian angle in AI compliance for the DACH market.

Does it reach a small practice in Waterford? Directly. Ireland is in the EU, so the Act covers a business here like any other in the Union. It also reaches further: providers placing AI systems on the EU market wherever they are based, and providers and deployers outside the EU where the output of the system is used in the EU.¹⁶

Worked example: which rules touch Ballybricken's three AI ideas

1. Staff using a chatbot to draft client emails. Not high risk under the AI Act. The question is data protection: which plan, which contract, and what staff are allowed to paste. The GDPR applies.
2. A tool that ranks applicants for a trainee bookkeeper post. This is the kind of use Annex III lists under employment, so it is high risk under the AI Act, with obligations from 2 December 2027.¹⁷ The GDPR rules on automated decisions and fairness apply on top.
3. A chatbot on ballybricken.example answering questions from clients in Waterford and Kilkenny. The practice deploys it in the EU, so the AI Act applies, and the transparency rules in Article 50 require AI systems that talk to people to be designed so that people know they are dealing with an AI, unless that is obvious.¹⁶

Three ideas, three sets of rules. "Is AI legal for us?" never has a one word answer.

Ireland: the AI Act applies directly, and the DPC watches the data

There is no separate Irish AI Act to wait for. The AI Act is an EU regulation with direct legal effect in Ireland, as in every member state. What Irish law still has to do is say who supervises and enforces it. The General Scheme of the Regulation of Artificial Intelligence Bill 2026, published by the Department of Enterprise, Tourism and Employment on 4 February 2026, does that with a distributed model: established sectoral regulators act as the competent authorities, and a new statutory body, Oifig Intleachta Shaorga na hÉireann, the AI Office of Ireland, coordinates them as the single point of contact.¹⁸ For personal data, the regulator that matters is the Data Protection Commission (DPC). Its guidance on AI, large language models and data protection, from July 2024, is the place to start, and it names this article's risk plainly: personal data that employees put into a third party AI tool can bring new security and data protection risks for the organisation.¹⁹

The reality, in my experience, is that enforcement is inconsistent. I once reported a significant data breach to the UK regulator, the ICO, with full details of the customer data exposed, and received a generic reply five months later asking whether the issue still existed. After I confirmed it did, I never heard from them again. That was in the UK, not with the DPC. The legal framework exists. Active enforcement is another matter.

That is no licence to ignore it. Enforcement is unpredictable, the damage to your reputation from a breach can be severe, and people whose data is exposed have their own right to compensation from the controller under Article 82 of the GDPR.¹² They do not need the regulator to act first.

Self hosting: what it solves and what it costs

Running models on your own infrastructure removes most of the privacy problems above. Data never leaves your network, and nothing you type trains someone else's model. For organisations with strict duties of confidentiality, such as healthcare, finance and legal services, it may be the only option they can defend.

Setting it up properly

Self hosting has become far easier. A basic setup has five parts:

1. Hardware. A server with a GPU sized to the model, worked out as in the memory example above.
2. A runtime. Ollama is the friendliest. vLLM or llama.cpp give more control.
3. A model. A small one (7 billion parameters) is fast and needs little hardware, a large one (70 billion) handles harder reasoning.
4. An API. Ollama exposes one automatically. By default it listens only on the machine itself, 127.0.0.1 on port 11434, and its FAQ shows how to change that and put it behind a proxy such as nginx.²⁰
5. Security. The server sits on its own network segment without internet access, behind a firewall, requires authentication and accepts connections only from approved systems. Keep logs of prompts and responses with a sensible retention period, so you have an audit trail. And staff still need clear rules on what may be processed.

A minimal check that the model answers, on the server itself:

ollama pull <model-name>
curl http://localhost:11434/api/generate -d '{"model": "<model-name>", "prompt": "Summarise the VAT cash receipts basis in two sentences."}'

The trade offs

Self hosting is not a magic answer. Whether an open weight model is as good as a cloud model for your task is something to test, not assume. Updates, patches and troubleshooting are yours, with no support ticket to open. The hardware is a real upfront cost, electricity adds up, and keeping it running needs skills your team may not have.

Worked example: Ballybricken's self hosted server, costed

Every figure here is the example's assumption. Eimear's developer prices a server with a GPU at €6,000, written off over three years. It draws an average of 300 watts around the clock, at an assumed €0.25 per kilowatt hour, and needs about two days of a contractor's time a year at €450 a day for updates.

Electricity first. A year has $24 \times 365 = 8{,}760$ hours:

$$0.3 \times 8{,}760 \times 0.25 = 657$$

Here $0.3$ is the draw in kilowatts, $8{,}760$ the hours in a year, $0.25$ the assumed price per kilowatt hour in euro, and $657$ the yearly electricity cost in euro.

Then the yearly total:

$$\frac{6{,}000}{3} + 657 + 2 \times 450 = 3{,}557$$

Here $\frac{6{,}000}{3}$ is the server cost spread over three years, $657$ the electricity, $2 \times 450$ the maintenance days, and $3{,}557$ the yearly total in euro.

So the server costs Ballybricken about €3,557 a year against €4,200 for the cloud seats. The money is close. The difference is what you get: the server keeps every client's payroll inside the building, but runs a smaller model and needs someone to look after it. Which is why a hybrid is worth a look.

The hybrid approach: route by sensitivity

A hybrid setup routes each request by how sensitive it is. Confidential data stays on the self hosted model. Requests with nothing sensitive in them, where the extra capability of a cloud model is worth having, go to the cloud.

Read it from the top. Every request passes the same check. Yes keeps sensitive content on your hardware, no sends it to the cloud, and both end in the log, which lets you show later what went where.

The check can be simple or clever. Simple versions match patterns (an email address, a PPS number, an IBAN, something shaped like an API key). Better ones add a small classification model that judges what kind of request it is. Some systems redact the personal data, send the cleaned request to the cloud and put the original values back into the answer.

A few tools make this practical. A gateway such as LiteLLM presents one API in front of many models, so your applications integrate once and the gateway decides where each call goes.²¹ Frameworks such as LangChain also support several model backends, so routing can live in your own code.

Worked example: a first router for Ballybricken

The simplest useful router is a few patterns and one rule. This sketch in Python is a starting point, not a complete filter:

import re

PATTERNS = {
    "email": re.compile(r"[\w.+-]+@[\w-]+\.[\w.]+"),
    "pps_number": re.compile(r"\b\d{7}[A-W][A-IW]?\b", re.I),
    "iban": re.compile(r"\bIE\d{2}\s?[A-Z]{4}\s?\d{4}\s?\d{4}\s?\d{4}\s?\d{2}\b", re.I),
    "api_key": re.compile(r"\b(sk|pk|api|key)[_-][A-Za-z0-9_-]{16,}\b", re.I),
}

def route(prompt: str) -> str:
    hits = [name for name, rx in PATTERNS.items() if rx.search(prompt)]
    if hits:
        return f"local (found: {', '.join(hits)})"
    return "cloud"

Run four of Ballybricken's real requests through it:

1. "Rewrite this paragraph of our newsletter so it sounds less stiff." Nothing found, so it goes to the cloud.
2. The payroll paste from earlier. It contains PPS numbers and an IBAN, so it stays local.
3. A developer's question with api_key = "sk_live_..." in the snippet. The key pattern matches, so it stays local, and the log tells Eimear a key was nearly sent out, which is worth a quiet word.
4. "Explain how VAT works on takeaway food for a client who runs a café." Nothing found, so it goes to the cloud.

The patterns will miss things (a client's name on its own, for example), which is why the policy and the training still matter. The router catches the obvious mistakes that people make on a busy Friday.

A hybrid in practice: GrowCentric

I've worked on exactly this kind of hybrid architecture for GrowCentric.ai, with a self hosted AI layer between the cloud AI and the user interface. When users interact with GrowCentric's guided input, their requests reach the self hosted component first.

That layer does three jobs. It processes sensitive data locally without exposing it to cloud systems. It prepares and enriches requests before they reach the more capable cloud models. And it filters the responses, so what comes back to the user has been checked.

The guided input means users are not typing free form prompts. They choose from structured options the system understands, which sharply reduces the risk of accidental exposure, because nobody is pasting raw data into a text field. The work that needs company data happens locally, and only cleaned, structured requests go on to the cloud when more capability is needed.

It is more work to build than either pure approach. For organisations with genuine privacy duties that also want the most capable models, it is the best balance I know.

Getting AI right in your organisation

One in five Irish enterprises with 10 or more staff now uses AI, and more than half of the large ones.² Some of your staff may be using it already, whether or not you said they may. The question is how to do it without risks you would never accept elsewhere. Cloud AI gives the most capability for the least effort. Self hosting solves the privacy problem at the price of hardware and upkeep. A hybrid routes the sensitive work home and sends the rest out.

Whichever you choose, five things hold:

1. Write a clear policy on what data may and may not go into which AI tool, and tell every member of staff.
2. Add technical controls that stop sensitive data reaching cloud AI, by routing, redacting or blocking.
3. Train your people. The best architecture does not help if staff do not know where the line is. In the EU it is also a legal duty to take measures that support it.¹⁷
4. Monitor use. Know who uses which system, for what, and whether the answers are fit to use.
5. Review it regularly. This field moves in months. A policy written today needs another look before the year is out.

AI can take tedious work off your people, surface what is hiding in your data and speed up decisions. Get the architecture right from the start and it becomes an advantage. Get it wrong and your clients' data ends up on someone else's servers, or in the news.

Footnotes

1. Central Statistics Office, Business in Ireland 2025: Digital sustainability and artificial intelligence adoption, CSO, 2026, https://www.cso.ie/en/releasesandpublications/ep/p-biistit/businessinireland2025-sustainabilitythroughinnovationandtechnology/digitalsustainabilityandartificialintelligenceadoption/

2. Central Statistics Office, Information Society Statistics, Enterprises 2025: Artificial Intelligence, CSO statistical release, 6 February 2026, https://www.cso.ie/en/releasesandpublications/ep/p-isse/informationsocietystatistics-enterprises2025/artificialintelligence/

3. Meta, Meta Llama 3 Community License Agreement, section 2, 18 April 2024, https://www.llama.com/llama3/license/

4. Mistral AI, Announcing Mistral 7B, 27 September 2023, https://mistral.ai/news/announcing-mistral-7b

5. DeepSeek, DeepSeek-R1 repository and licence (MIT), GitHub, 2025, https://github.com/deepseek-ai/DeepSeek-R1

6. Hugging Face, Quantization overview, Transformers documentation, accessed November 2026, https://huggingface.co/docs/transformers/main/en/quantization/overview

7. Anthropic, Updates to Consumer Terms and Privacy Policy, 28 August 2025, https://www.anthropic.com/news/updates-to-our-consumer-terms

8. Arjun Kharpal, Samsung bans use of generative AI tools like ChatGPT after April internal data leak, CNBC, 2 May 2023, https://www.cnbc.com/2023/05/02/samsung-bans-use-of-ai-like-chatgpt-for-staff-after-misuse-of-chatbot.html

9. NOS Nieuws, Privacywaakhond: pas op met gebruik AI-chatbots, ging mis bij huisarts, 7 August 2024, https://nos.nl/l/2532234

10. Accountant.nl, AP slaat alarm over datalekken door AI, 30 December 2025, https://www.accountant.nl/nieuws/2025/12/ap-slaat-alarm-over-datalekken-door-ai/

11. Lobke Kapteijns, Datalek: gemeente Eindhoven deelt persoonsgegevens met openbare AI-websites, Omroep Brabant, 19 December 2025, https://www.omroepbrabant.nl/nieuws/6001041/datalek-gemeente-eindhoven-deelt-persoonsgegevens-met-openbare-ai-websites

12. Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 6, 28, 44, 82 and 83(5), EUR-Lex, https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32016R0679

13. Commission Implementing Decision (EU) 2023/1795 on the adequate level of protection of personal data under the EU-US Data Privacy Framework, 10 July 2023, EUR-Lex, https://eur-lex.europa.eu/eli/dec_impl/2023/1795/oj/eng

14. General Court of the European Union, Press release No 106/25, Judgment in Case T-553/23 Latombe v Commission, 3 September 2025, https://curia.europa.eu/jcms/upload/docs/application/pdf/2025-09/cp250106en.pdf

15. Garante per la protezione dei dati personali, ChatGPT, il Garante privacy chiude l'istruttoria, 20 December 2024, with the note on the Rome court ruling published 18 March 2026, https://www.garanteprivacy.it/home/docweb/-/docweb-display/docweb/10085432

16. Regulation (EU) 2024/1689 (Artificial Intelligence Act), Articles 2, 4, 50 and 113 and Annex III, OJ L 12 July 2024, EUR-Lex, https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689

17. Regulation (EU) 2026/1744 (Digital Omnibus on AI), amending Regulation (EU) 2024/1689, 8 July 2026, OJ L 24 July 2026, EUR-Lex, https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202601744

18. Department of Enterprise, Tourism and Employment, General Scheme of the Regulation of Artificial Intelligence Bill 2026, 4 February 2026, https://enterprise.gov.ie/en/legislation/general-scheme-of-the-regulation-of-artificial-intelligence-bill-2026.html

19. Data Protection Commission, AI, Large Language Models and Data Protection, 18 July 2024, https://www.dataprotection.ie/en/dpc-guidance/blogs/AI-LLMs-and-Data-Protection

20. Ollama, FAQ: how do I expose Ollama on my network, accessed November 2026, https://docs.ollama.com/faq

21. LiteLLM documentation, accessed November 2026, https://docs.litellm.ai/docs/