This page is a translation. The page of record is the Portuguese one, and every claim on it walks back to the bytes named there. Portuguese original · 98% translated

NoticeMethodAPINewsroom ENPTENFRDEv0.23.13
Lisboa · última captura: segunda-feira, 14 de setembro de 2026
Startup Summit Lisbon 2026 · 17–18 de setembro de 2026

O Ecossistema Português de IAbeta

Portugal's AI ecosystem is small enough to be mapped completely and large enough to be interesting.
Data protection notice

What this publication holds about named people, why, and how to ask to be taken out — giving no reason at all.

This notice is what a publication owes the people it names, written by whoever publishes. It is not legal advice and no lawyer has reviewed it. The article it rests on — Article 24 of Law no. 58/2019 — came back truncated to the research that found it, and must be read in full before being cited on a public page. If you are using this as a model, seek your own advice.

The order in which this was published

This notice was written and published in the SAME version as the first pages that name people, and before any of them was online. The /portugal/ section of newsroom.sgit.ai, where this method comes from, published its notice AFTER the pages that named 64 people, and records on the page itself that this was the wrong order. The commissioning brief expressly requires that this mistake not be repeated. It is the only reason this file exists before a single page of people does.

Controller

Dinis Cruz, in a personal capacity

There is no entity behind this publication. Naming a person rather than a project is the honest version and it is also the cheap one: it means the objection route below reaches somebody who can act on it, and not a form.

Contacto: dinis.cruz@owasp.org

Who is covered

People named in the speaker list published by the event itself, in their professional capacity: the name, the role they are listed under, the organisation they are listed under, the links the event published for them, the topics the event lists on each of their pages, and the words on that page that match a published lexicon.

Finalidade

To report on a public professional event and on the ecosystem around it, and to demonstrate a method in which every published claim can be walked back to a hashed copy of the source it came from.

What is held
  • Name, as published by the event
  • Professional role, as published by the event
  • Organisation, as published by the event
  • The address of that person's page on the event's site
  • A LinkedIn profile address, where the event published one
  • Os tópicos que o evento lista na página da própria pessoa, verbatim
  • The words on that page that match a published lexicon of sectors, technologies, ideas, services and products — only the matched words, never the sentence
What is refused
  • Postal addresses of any natural person
  • Email addresses of any natural person
  • Telephone numbers of any natural person
  • Any special category data
  • Biographies as text: they are the text of whoever wrote them, and they are linked rather than reproduced. A published lexicon reads them looking for words and keeps only the words
  • Any assessment, ranking, scoring or characterisation of a named person. A derived tag says the page contains certain words; the connections between entities are made at organisation level and name nobody

The refusal runs at the point of READING, not of presentation: build/extract.py discards the field before it reaches a data file. The gates in build/gates.py are the second line — gate 5 fails the build if an email address, a telephone number or a natural person's postal address appears in any data file; gate 6 fails it if any field is long enough to be a reproduced biography; gate 8 runs every lexicon pattern over the frozen bytes again, so that a tag can never be written by hand. A contact detail that reaches the data and is merely hidden by a template is one distracted cycle away from being published.

Lawful basis · Legitimate interests

The journalistic derogation is NOT invoked. Article 24(3) of Law no. 58/2019 conditions processing for journalistic purposes on the national legislation governing access to and exercise of the profession. This publication is not accredited under it and does not satisfy that condition, so it does not rely on it. The supervisory authority's 2019 decision not to apply certain provisions does not touch Article 24, which therefore stands in full.

The interest

To report what a public event published about itself, and to demonstrate a method of verification on top of that. Both are matters of public interest and neither is commercial — this site has no price, no advertising and no tracking.

The need

The account cannot be given without the names, because the subject IS who is speaking. The data held is the minimum: the published professional fields and nothing else. No biography is kept, no contact detail is kept, and no inference about anybody is made or published.

The balance

Every datum was published by the event itself for the express purpose of announcing these people as speakers, in their professional capacity, and holds no private-life content. Against that: being listed on a third party's site is not something anybody consented to, and that is why the objection route below is unconditional rather than assessed. We also record, because otherwise it would go unsaid and the wrong belief would be the one left standing, that the fact data has been made public does NOT by itself mean it may be processed on this basis — it is a factor in the balance, not an answer to it.

The three members follow the European Data Protection Board's guidelines on legitimate interests, adopted on 8 October 2024.

Why a public notice

Where personal data is not obtained from the person it concerns, Article 14 of the GDPR requires that they be informed. Article 14(5)(b) releases the controller where individual notification would take disproportionate effort — but that release is conditional, and one of the measures it expressly requires is making the information publicly available.

This page is that measure. It exists so that the release is earned and not presumed. It was written before a single page of this site named a person, and not after. It is the correction the commissioning brief requires to be carried over from the section that came before this one.

Your rights
  • Ask what is held about you, and get a copy
  • Correct anything inaccurate
  • Object to being included, and be removed
  • That the removal propagates to every page on the next build, which runs on request and not only on a schedule
  • Complain to the Comissão Nacional de Proteção de Dados, or to the supervisory authority of the country where you live
Objection and removal

Como: Email the controller named above. Say who you are and that you want your entry removed.

No reason is required and none will be asked for. Removal is not assessed, discussed or deferred pending a balancing exercise.

The entry is removed from the derived data, the site is rebuilt, and the name stops appearing on any page. Because removal changes what a difference between captures would show, the changes page records that an entry was removed at the data subject's request, without a name — which is the only way to keep the difference honest without republishing the name somebody asked to have removed.

When a name leaves a list

When a name left the list published by the event itself between two of our captures, this publication records the change and declares no reason. Withdrawal, a diary clash, a duplicate registration and an editing error are indistinguishable from outside, and attributing a motive to a named person on this evidence would be the most harmful thing this publication could do.

What this site does not do
Retention

The frozen captures of the sources are permanent, because they are the evidence that makes every claim verifiable and deleting them would break the method. The derived data is rebuilt from them on every run. If a person objects, their entry is removed from the derived data and from every page presented; the frozen capture is not altered, because it is a copy of what somebody else published on a date, and not our record about them.

Masthead

Editor of record: Dinis Cruz. Research, writing and verification: agents, each one named in the run record that did the work.

Sources

All fetched, frozen byte for byte and hash-verified on every build. The register lists them; the method explica-o.

Data protection

Controller: Dinis Cruz. Basis: legitimate interests. Unconditional removal on request; the reason is never published. The notice.

Part of

newsroom.sgit.ai — one instance of the argument. This site is natively Portuguese; it is not a translation.

The articles · Method · The newsroom · Research deliveries · The desk · The wallet · The files · Data protection notice · About and limits · Newsroom (EN) · Versions
For an agent. This page is generated from JSON files in dados/ by build/build.py; nothing on it is written by hand. The machine-readable index is at llms.txt and the manifest with the hash of every data file at ficheiros/. Version v0.23.13 · source: github.com/SGit-AI/SGit-AI__Website__Newsroom__PT · file: aviso/index.html.