What’s new
Every change to the Memorial, the map, the Legislative Observatory, the Media Observatory and the site, from most recent to oldest. Turing versions have their own section.
Cuando una publicación es parte de un hilo de la misma cuenta, Turing lo lee y lo califica completo, como una sola nota.
Lo que dijo cada persona legisladora ahora se ve con la forma de donde salió, igual que en la página de cada nota.
Turing ahora es un chip azul. Cuando piensa, sus pistas se encienden con los colores de la bandera, y en la barra de arriba ya tiene su botón.
En Recursos ya están las jurisprudencias y sentencias de la Corte sobre matrimonio igualitario, identidad de género, infancias trans y terapias de conversión, con el enlace oficial de cada una.
If all we can read of an article or post is the headline, a short snippet or the Facebook preview, Turing does not rate it: it asks you to paste the full text and rates it as soon as it has it.
En «Dónde lo encontramos», una franja ámbar con ✋ marca la parte de la nota que va contra los derechos LGBTTTIQANB+, y la página quedó mejor ordenada.
The usurpation or simulation of an affirmative action now counts as news that goes against our rights. Signs of possible simulation lead to Turing’s analysis of the article, and each legislator’s profile shows its related articles.
On each legislator’s profile, the “signs of possible simulation” box appeared in pink with white text. It is now dark magenta with light text, on computers and phones.
When Turing could only read a social media preview, the headline and summary, or the beginning of a document, the result includes a box to paste the full text. Turing grades it again with all its steps, in its usual window, and the rating is no longer preliminary.
The 0 to 100 rating measures the writing. Now, next to “The essentials”, Turing also tells you whether the news is bad for the community (red), goes against our rights (yellow), is good news (green) or just informs (blue).
Turing keeps watch over the posts it analyzed. If the account changes the text, it saves the new version, certifies it with another folio and marks it on the article, the certificate and the legislator’s profile. For their evaluation, the worst of their versions counts: correcting is acknowledged, but it does not erase what they said.
Before you download the copy (manifiesto.json) or the seals (.tsr and .ots), we show you how to check them yourself, with the commands ready to copy. The PDF is in red and each file has its own color.
On each certified copy, the PDF, the copy (manifiesto.json) and the timestamps (.tsr and .ots) are in a box on the right that scrolls with you. On mobile they open with the «Download» button without moving the page.
Every page of a certified copy's PDF carries the same header with the logo, the folio number and the QR code, and paragraphs are no longer split between pages.
En la página de Turing, «Cómo funciona» es más ancho y suma el paso 5, la copia certificada: huella digital, sello de tiempo de un tercero, registro en Bitcoin, copia en Internet Archive y vigilancia por si la borran.
En la página de cada nota analizada, el texto con sus subrayados aparece arriba, justo después de la calificación, porque es lo que Turing leyó y calificó.
The certificate page and its PDF carry our logo at the top, the title «CERTIFIED COPY» and the gold seal on the left.
El certificado de cada copia certificada ahora es una hoja blanca y elegante, pensada para imprimirse, con un sello dorado que lleva el logo de Memoria y Resistencia.
Cada nota y publicación analizada por Turing muestra su copia certificada como una constancia formal: folio, código QR, cadena de custodia y qué valor tiene como prueba.
During the Ofrenda season, the background of each person's page and its papel picado carry the colors of their flag.
The Memorial and case pages leave purple behind: they are ivory all year and, from October 15 to November 15, take on an ofrenda style with cempasúchil marigolds, Mexican pink and papel picado.
We added the “Use of Turing” section: what data it uses, what it never shares with other companies and what you can request.
The PDFs for each analyzed article, each case and each legislator now say when there is a certified copy of the publication and whether it was deleted later.
En la ficha de un caso, la ventana del aviso de una fuente y la de la foto ampliada ahora usan el ancho de la página y se acomodan solas en compu y celular.
Si una persona borra después lo que publicó, se puede demostrar que existió: Turing guarda su copia con huella digital y sello de tiempo de un tercero, y la manda a Internet Archive.
El aviso de donar es ahora una pastilla pequeña en azul y morado que se esconde cuando abres una ventana.
Turing no longer just rates language: it fact-checks, connects each article with the whole site and learns. Because of that fundamental change it goes from 1.6.6 to 2.0, and from now on its versions are numbered with clear rules.
La herramienta para analizar notas vive en memoriayresistencia.org/medios/turing y usa el azul de Turing. Los enlaces viejos a /medios/turing siguen funcionando y te llevan ahí.
Turing ahora distingue si la persona de una nota sigue con vida, murió o está desaparecida. Las agresiones a personas que sobrevivieron se suman a una estadística nueva, con las palabras que usó la prensa para contarlas.
La página para donar ahora mezcla el azul de Turing con el morado del sitio, y los encabezados de las páginas nuevas tienen su propia animación.
En el Analizador de notas, la sección «Cómo funciona Turing» ahora muestra el video «Te presentamos a Turing».
El buscador del sitio ahora trabaja con Turing: reconoce de quién o de qué hablas, te muestra lo relacionado y ordena los resultados para que lo importante salga primero.
En Entrenar a Turing, las fotos ya no se ven como un cuadro borroso: salen en un marco con un aviso claro y, al abrirlas, se ven completas y más nítidas.
La portada del Observatorio, la página de cada congreso y la ficha de cada iniciativa estrenan encabezado: lo importante se ve de un vistazo. Y ahora te decimos qué partido es realmente aliado.
The Open Data page has a new look and the Related report now has search boxes to choose the period, a state or which states to show in the table.
Cuando una actualización trae algo nuevo que no está a la vista, ahora te lo decimos y, con «Iniciar presentación», la página baja sola hasta ahí. También puedes cerrar la presentación.
El motor del Analizador ahora se llama Turing, en honor a Alan Turing. Ya no es solo el motor de una herramienta: es el sistema propio de Memoria y Resistencia que analiza, decide y aprende con la información de todo el sitio.
La ficha de diputaciones y senadurías se rediseñó para entenderla de un vistazo: arriba dice su postura y sus cifras clave, y un menú te lleva a cada parte.
Ya no hay un Verificador aparte: cuando analizas una nota, Turing también compara sus cifras, leyes y casos —y lo que dicen las organizaciones y personas legisladoras que cita— con el Memorial y el Observatorio Legislativo.
The home page alerts no longer send you straight to the outlet: they open what the Analyzer found, and the link to the original article warns that it is on another site.
A single report that cross-references the Memorial, the Media Observatory and the Legislative Observatory: how many people we lost, how the press told it and what congresses did in the same period.
Besides reviewing how it is written, each article’s page compares its figures, laws and cases (and what the organizations, legislators and authorities it quotes say) with the Memorial, the Legislative Observatory and published fact-checks.
As you scroll down a bill’s page, the short version of the traffic light appears at the bottom right, just above the donations button, without covering it.
The versions in other languages now include everything new: the Legislative Observatory, the shams, the search, the news page and the Analyzer’s messages.
Did you get a figure, a news story or a chain message? Paste the fact, the link or a screenshot and we’ll compare it with the Memorial, the Legislative Observatory and fact-checks already published.
When Turing improves, the page of an article analyzed earlier tells you which version rated it and brings it up to date with a button.
How we assess congresses and each legislator now lives in Resources, next to the citizen guides and the library.
In the Legislative Observatory you can now report, with a press article, anyone who holds a seat reserved by affirmative action without belonging to the group. With three articles from different outlets well rated by the Media Analyzer, the report is accepted and the person’s record shows it.
Each record has buttons for the person’s official accounts on X, Facebook, Instagram, TikTok, YouTube and Threads, and the link to their record on the congress website.
The end of each page is now lighter: fewer links and the important things in view.
The record, its PDF, the lists, the parties and each bill use the same rules to say who is an ally, who isn’t, who fakes support and who holds back the agenda.
The menu’s magnifying glass now opens a window: if you type, it searches the whole site; if you paste the link of an article or post, it sends it straight to the Analyzer.
The memory record and an article’s analysis as PDF are built on the spot with the same thing you see on the page, with no saved copies.
A new page gathers every change to the site, from the most recent to the oldest, and the home page debuts the “New” section with the latest from the whole project.
When the site shows you what changed, the invitation to donate waits until you finish and appears a few seconds later.
The PDF of a deputy’s or senator’s record is built on the spot with the same thing you see on their page: their photo, their labels and what they have said in the media.
When a statement doesn’t pass all the checks, the third AI review decides it even if only one AI has quota at that moment, and meanwhile the article shows what the person said.
What legislators say to postpone, evade or downplay our rights appears on their record and on the bill, and counts in their evaluation.
The Observatory recognizes legislators even if the article names them by their middle name or only by their surnames, and reads the verbatim quote even if it comes paragraphs after their name.
When a proposal in favor of our rights is still in the freezer and most of what is said about it in public is against, the Legislative Observatory warns that it may be a sham: introducing it for appearances and keeping it frozen so it isn’t approved.
In the Legislative Observatory, being an ally requires introducing proposals or speaking out in favor in public. Those who only signed appear separately, and those who sign in favor but speak against are marked “Fakes support”.
When a legislator replies to someone on X, the Analyzer reads the whole conversation to understand what topic it’s about and records at once everything that person said in the same thread.
If a legislator writes their position in the first person and plainly from their official account (“I am against trans childhoods”), the Analyzer records it on their record even if the artificial intelligence is overloaded.
When a legislator replies to someone on X, the Analyzer checks their profile with the data that comes in the same post, so it knows whether it is their official account even if X is slow to respond.
When a legislator posts from their own account, we check online whether it is their official profile. If it is, what they say there counts as an official source on their record.
Now the Analyzer also reads Facebook posts and recognizes the legislator by the name of their account or page. What it finds goes to their record without waiting for the team, and the Observatory’s “Analyze a statement” uses the same Analyzer.
When you analyze an article or a post, the same progress box shows you each part: the language, whether it names someone in the Memorial, the legislators and what they said, and whether other outlets confirm the statement. When it’s done, the article’s page opens with everything ready.
The essentials are at the top: the rating, the central idea and how many things are worth changing or done well. An index goes with you as you read so you can jump to each part.
When the Analyzer finishes, it takes you to the article’s public page, with its link ready to copy or share on X, Facebook, Threads and WhatsApp.
The statements you send through “Analyze a statement” no longer ask for your email: they are saved in “My statements”, inside your account, and there you see when each one is confirmed.
If you paste several links together, none of them gets lost anymore.
If you paste an article’s text, the Analyzer rates that text even if you press “Analyze” next to the link. And saved analyses are brought up to date with the current Turing when you open them.
When analyzing an article, the AI decides right then whether it talks about a bill and what position each legislator shows, without leaving it on hold.
If an article says a party or its leadership avoids taking a position, the Legislative Observatory box shows it in “Other voices”, even if the article doesn’t say which congress it is about.
When you analyze an X post, the AI listens to what is said in its video and reads the article it links. With that it locates the topic, the congress and the bill being discussed.
It no longer corrects the gender of people unrelated to the case: it only checks how the article names the trans person.
Groq stopped offering the Llama model for free, so the second opinion that comes through Groq is now given by OpenAI’s GPT-OSS. Gemini is also back, after being left out by an error.
Now an artificial intelligence helps review comments and requests from outlets. What can’t be undone, such as replying to an outlet, is still done by a person on the team.
The Library adds the OSIEGCS Protocol, the Notes on prejudice-motivated crimes and the Supreme Court’s Case Law Notebook, plus Booklet 19 of the Inter-American Court, each with its reading sheet.
The article Analyzer now reviews news with the criteria of the Supreme Court and the Inter-American Court on how to name and how to report violence against LGBTTTIQANB+ people.
The daily search now also looks for negative press and warns when an article reports a possible rollback in the rights of LGBTTTIQANB+ people or on issues such as abortion, disability or HIV.
When a statement’s verification falls short, a third AI review decides within minutes whether it is published, without waiting for the team, which can correct any position.
When an update arrives, the page dims for a moment, highlights each new section or button and tells you in one sentence what it’s about.
You paste the link and the AI does the rest: it finds who said what, checks it against other news and only asks you what it can’t solve on its own.
The Analyzer shows the logos of the AIs it uses and lights up in color the one that reviewed your article.
Now you can paste the link of an article or post on the record of a deputy, senator or bill, and the site searches on its own, every week, for what they say about LGBTTTIQANB+ topics.
For a statement to count in a legislator’s evaluation, it must now be confirmed by an official source or two independent outlets.
The second opinion from artificial intelligence no longer depends on a single service: each review is done by the AI that works best for that job and has quota that day.
The home page shows the latest videos from our YouTube channel and the project’s four paths: the map, the Memorial and the two observatories.
In the Media Observatory you can see the articles about LGBTTTIQANB+ people for any day with a calendar.
Go see itYou can read Memoria y Resistencia in Spanish, English, Portuguese and French. The language is chosen automatically according to your country and you can change it at the top.
Comments on each case’s record have clear rules. Hate messages are not published and any sanction can be appealed.
Go see itA new page shows on which topics the press uses the most stigmatizing language about LGBTTTIQANB+ people.
Go see itEach section of the site has a header with its own light animation that respects those who prefer less motion.
Cases, comparisons, “create your chart”, media, stigma by topic and congresses: all ready for research, journalism and advocacy.
Go see itA daily search reviews the press and adds the articles on the topic to the Media Observatory. Possible cases are reviewed by a team member before being published.
Go see itThe home page summarizes what is pending in the Chamber of Deputies, the Senate and the state congresses: what moves forward, what is frozen and who delivers results.
Go see itPaste the link to an article or a post and find out how it names LGBTTTIQANB+ people. Each article on the topic is added to the Media Observatory.
Go see itThe sources of each case are shown in cards, you can download an image to share on social media and, from your phone, register a case step by step.
Go see itA new page goes through the cases over time, and the data can be downloaded as CSV or JSON.
Go see itA clearer, faster site made to read well on your phone.
There is nothing new in this area.
