Rocío Santamaría | Consultora SEO

SEO Estratégico & AI Search | Posicionamiento web + Marketing Digital para Tiendas online, servicios y negocios locales

[Rocío Santamaría | Consultora SEO @RocioSantMK]

🥳 ¡Inauguro nueva sección en mi web!

He decidido recapitular una serie de casos de éxito de proyectos reales. Desde estrategias SEO para tiendas online, blogs y webs de servicios hasta negocios locales.

Acabo de estrenarla con un caso de éxito de SEO local.

[Alex Cui @alexcdot]

Claude's watermark probably doesn't work how you think. As the CTO of GPTZero, I'll explain how Anthropic, Google and OpenAI are building text watermarking in this brief explainer and whether it can be defeated.

Almost all forms of watermarking that are fast and cheap enough for a frontier lab have the same formula, following the KGW method:

In generation:

1. Let's say you've generated n tokens so far. Take those n tokens + a secret key to generate a random hash
2. Use that hash to randomly reweight the probabilities for the n+1 token, and then sample from that new distribution. In the simple case, you could split 50% of all English words into a green or red set based on your hash, and boost the probability of words in the green set.

For watermark detection:

1. For each token, see if it was in the green or red set.
2. To do this, recreate the hash based on the secret key and the text preceding the current token. Then, recreate the green and red set of words.
3. Once you've checked all the words in the text, if the next token is selected disproportionally from the green set more than 50% of the time, you claim the text has the watermark.

I can tell you want to ask the following:

1) Isn't it easy to mess up the hash if you paraphrase the text? The answer is mostly yes, however, you can use a statistical model to get your hash instead of a deterministic function (SIR, Adaptive Watermark). Since the entire watermark is probabilistic, this is fine.

2) Doesn't this make the text much worse? The answer is yes, it does - Yes, it does – but for most people, it's imperceptible (Google claims in human feedback study with 20,000 texts), since there are exponentially many ways to write the same paragraph. DiPmark does something more sophisticated to avoid shifting the text distribution on average. Of course, watermarks fail on short text or highly predictable texts like "2+2=4".

3) Shouldn't it be easy to figure out the green and red sets? The answer is no. You would need an exponentially large number of samples from the watermarker to reconstruct those sets exactly, but it's a risk if the detector is open to the wild (Watermark Stealing)

Still, there are couple challenges that a frontier lab needs to overcome:
1. Their watermark needs to work token-by-token because they are streaming their text to users. Many watermark methods plan sentences or paragraphs at a time, or change the text after its entirely written, in order to make their watermark robust to paraphrasers, and a frontier lab cannot afford to do this yet (SemStamp, PostMark)
2. If the secret key leaks, the watermark is busted. To avoid a large blast damage from this, you need to have a couple secret keys in rotation.
3. There are some texts, like code, that cannot be arbitrarily changed, otherwise the code will break. In those cases, the watermark needs to selectively change words in parts of the text that can tolerate synonyms (i.e. like variable naming) - see SWEET, EWD, Invisible Entropy.
4. They will need to educate their users on how to deal with false positives and false negatives of a detector, which is a big challenge (one we put a lot of effort into)

So, how do I see this playing out in the next 6 months?
1. If Anthropic releases the watermark detector publically, I think they defeat their own watermark. People find reliable watermark removal strategies by testing against Anthropic (AI detectors like GPTZero have an advantage here because they can train against these adversaries once they become popular).
2. If they keep the detector private to the government, like Google has done, it's "safer". However, there are some papers showing trained approaches that work robustly to zero-shot break watermarks without any data, simply because they try to write the text just like a human (Zhang et al. 2024, Watermarks in the Sand). Also, making your detector makes it battle-tested and stronger long-term (my experience).
3. In my testing, the watermarks don't survive intense paraphrasing (especially if you combine word choice and syntax attacks), or human text substitution (rewrite your AI text by plagiarizing human authors). The free paraphrasers I've tried have quickly bypassed Google Deepmind's SynthId for what it's worth.
4. All-in-all, frontier labs are likely okay with this because they expect most users to not attack the watermark, and also because they + European regulators likely don't care past a certain point - its good enough.
5. Overall, I think users of frontier LLMs will not really care about this, because 1) they don't realize watermarks are there, 2) EU will force everyone to conform, 3) this seems more like regulatory hoop-jumping than an earnest effort from frontier labs to expose LLM use

Lastly, people's first concern shouldn't be watermarking, it should be AI detectors!

If you're posting, "its not X, its Y!!", I don't think the watermark is going to make a difference :)

[Fernando Tellado Ⓦ @fernandot]

Si utilizas WPForms y otros plugins de Awesome Motive esto te interesa.

Denuncian que en la versión 2.0 habría introducido una puerta trasera con la que podría instalar cualquier cosa en tu web 💀

Si se confirma sería motivo de sobra para echarles del directorio de plugins 🫣

[AESAN @AESAN_gob_es]

⚠️ Alerta por posible presencia de cuerpos extraños (fragmentos de vidrio) en confituras y en miel .
🚫 No consumir.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@Alexserramar Felicidades, Alex ❤️

[Rocío Santamaría | Consultora SEO @RocioSantMK]

Y avísalo también en tus redes y en la web, que nunca sobra ;)

Es una tontería que te ahorra problemas.

¿Tienes ya actualizado tu horario de agosto en Google Business Profile?

[Rocío Santamaría | Consultora SEO @RocioSantMK]

El otro día un neumático de mi coche empezó a perder presión.

Busqué un taller cerca de casa, miré en Google y aparecía abierto. Llamé para acercarme y me dijeron que estaban de vacaciones hasta el día 10.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

- Si cierras una o varias semanas lo ideal es usar "cerrado temporalmente" y volver a abrir en cuanto regreses. Ponte una alarma para reactivarlo en cuanto vuelvas a abrir.

[David Bonilla @david_bonilla]

La historia detrás de la caída de la web de @MovieNumbers es de obligada lectura para cualquiera que se dedique profesionalmente a construir webs en la era de la IA. https://t.co/1oQ4DPEzXi

[MJ Cachón @mjcachon]

Google actualiza su guía de optimización para IA, ya mencionando expresamente GEO/AEO

https://t.co/zs94waghC6

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@andmorale Sí, cambiaré a Brave seguramente.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

Siempre he leído que los Mac raramente se pillan y que no es habitual ver la famosa pelota de colorines. Me compré hace unos meses el M5 pro con 48 GB de RAM y salta la pelota así siempre 2-3 segundos cuando uso Claude 🫣

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@jesush2o Todo normal y va bastante sobrado. Será el maldito Chrome.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@EstebanCervi Uso la aplicación de escritorio generalmente, tengo las carpetas y demás montadas ahí, es más fácil acceder. La terminal la dejo más para crear tools y cosas así.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@fernandot Al revisar el monitor tanto en CPU como Memoria va muy sobrado, vamos que el portátil ni se inmuta. Lo que más consume es Chrome, cambiaré de navegador.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@ServandoCPA Uso ambos, me da problemas con el navegador. Supongo que es Chrome.

[Rocío Santamaría | Consultora SEO @RocioSantMK]

@fernandot No escogí el más potente, el chip superior es el M5 Max. No se calienta, creo que tiene que ver más con Chrome y/o alguna extensión. Tengo que probar con otros navegadores.