The target audience for this particular rule of the AI Act isn't the producers or users of AI. It is the consumers of content that might be AI generated.

And frankly, letting people know if they see AI content or not is a fair thing to ask.

Fun fact: The EU AI Act considers this. It makes a difference between "AI generated" and "AI modified".

But it's hard to put this into the type of watermarking they are using. Because if you generate a text and then change a few things here and there what you are doing is degrading the watermark. But since the watermark is probabilistic, it's hard to tell if it was a weak watermark from the start or if it was degraded.

Stop changing the clocks twice per year, but if there are reasons to set opening or closing times earlier or later, I am all for it. Sunlight at whatever-oclock isn't bright enough? Then start school at whatever-thirty or other appropriate time.

The study could have been framed "it would be best to wake up later in the day" rather than "we should make the clocks read different numbers so that waking up at this particular number is better for us".

The problem is not that specific number correspond to particular positions of the sun, but rather than problem is that particular activities are tied to particular numbers on the clock. Framed this way, the solution is obvious, change the activities so they occur at other numbers on the clock.

("It actually makes RAM needs worse" -> "MoEs actually make RAM needs worse")

It actually makes RAM needs worse than a dense model, for a given quality.

That said, inference techniques like FreeToken are improving MoE swapping performance, so it's not as bad as it once was. Honestly, I would not be at all surprised if we start doing training in a swap-aware manner, where expert cache misses count as loss and models can queue experts to start loading before they're needed. Could even train in two tiers of loading - RAM and disk. Wouldn't that be great? Could run multi-terabyte models on consumer hardware. Training would tend to tend to concentrate key logic and reasoning on small number of very active experts, mixed reasoning/knowledge on less common experts that are usually kept in RAM, and rarer knowledge on experts that usually remain on-disk until needed.

There are well known, reputable abliterators on HuggingFace.

Also, censorship usually (not always) is something you care about for chats, not agentic work.

A .safetensors or .gguf model cannot run code unless you let it. No agentic framework = all it can do is write text. It can't hurt you directly. At worst, it can only lie to you.

It's big, but not good for its size. They're using tricks to pretend that they're better than they are. For example, compare the numbers that they list for the competion on DeepSWE up against the actual DeepSWE scores.

Not cool, Mistral.

Artificial Analysis is overrated, but yes, Mistral is playing fast and loose with their claims.

It takes one hell of a toll to try to do verbal exams 1:1 on students to evaluate them.

Nobody said "verbal exams 1:1".

The only thing that is required is that they be separated from their computers and their phones.

Nobody, I repeat, nobody who uses AI is going to see "OpenAI adding watermarks" and think, "Oh, I better use AI more, especially OpenAI".

And nobody, I repeat, nobody who doesn't use AI is going to see "OpenAI adding watermarks" and think, "Oh, I better use AI more, now that it'll be easier to see that what I did is made by AI."

It is not "an advertisement of the power of the technology". It's a regulation forced on them by the EU's AI Act, which neither companies nor users want.

"Guarantee" isn't what matters here. Likelihood is what matters. They provide predictions better than random chance. That's the central point.

Perhaps you should be realizing by now that homework is no longer a viable way to test how well students actually know the subject matter, and the only viable way to do so is in-class tests/exercises - frequent enough and worth enough of the grade to force them to study between lessons.

OR, alternatively, you could just keep doing what you're doing, and instead, in effect grade people on:

Comparably poor grades:
* People who actually do the work by hand
* "AI N00bs" who let stuff like that slip through

Comparably good grades:
* AI-experienced people who know enough to skim over the work, do multiple passes, or issue prompts that prevent such tells from getting into the work

Wherein, in effect, your net reward/punishment is inverted vs. what you should want it to be.

Closing one's eyes and ears to the fact that the environment students operate in has changed isn't an acceptable option. Education must adapt. And watermarks, BTW, will only catch "the n00bs"; more experienced / up to date students will keep track of what watermarks and what doesn't, and use AIs that don't watermark, or de-watermarking services.

I'll repeat: watermarks are anti-marketing. They discourage use.

SynthID in text genuinely is invisible. The probability shifts are small, and all the pathways are valid pathways to express the same thought.

And no, you cannot "collide" text watermarks. At best, it'll only have the latter one's watermark. At worst, both.

It'll still get tagged by SynthID.

You have to use a non-watermarking LLM. Saying "Don't do anything that will watermark it" doesn't help.