The question isn't only "should we" regulate the companies, but also "how"?
The current Trump Admin can't even decide on the first question, let alone come up for a plan on the second.
The Trump Admin's EO was to ask nicely all of the AI companies to give them 30 days to voluntarily review each model before wide release, but they have also failed to do that for the Mythos/Fable release, only to get a call from Amazon's CEO to David Sachs to convince them to disable Fable (to all non-US citizens).
We elected the party of "minimum regulations" into all 3 branches of government and we will reap what we sewed.
> There’s no need for an LLM, just a phd in virology, a decent lab, and a handful of grad students.
You are missing the forest for the trees. The existing virologist PhDs and the GoF labs are a scarce resource. The model makes the same scarce knowledge accessible to many more malicious actors.
The five eyes have four other eyes. But, from where I'm sitting the primary eye has only gotten stronger in the last 2 years. See some work we did in late 00's to early 2010's called Nexus 7 and extrapolate to modern capabilities.
I feel like your comment is relying heavily on bad heuristics (the entities capable of using LLMs to create weapons aren't all individuals) and you are making assumptions that the same "strictly controlled" regimes are in place.
It's worth verifying whether the US, OECD countries, UN, etc will have sufficient regulation over suspicious purchases, for example, after DOGE and funding cuts. This will be an ongoing issue as sovereign debts and bond yields squeeze out spending for other government regulation.
The weapons of mass destruction trio (bio, chem, nuke) have long been a concern for world powers.
Nuke requires a long supply chain and massive resources, so the worry there is maintaining control of all of the existing nuke weapons. Except for Russia racing towards Turin no itself into a failed state by staying engaged in the war with Ukraine, I don't see the nuke equation has changed much in recent years. Perhaps N Korea is a worry, but they seem to just want attention and power. Iran pretends to have nukes and says they want to extinguish Israel and the US, but I interpret that as posturing to maintain domestic control and Israel seems excellent at countering Iran's threats.
Chemical weapons have traditionally been the easiest to create (like creating chlorine gas from mixing common household cleaners). The trick there has always been volume and how to disperse it. I suspect within countries, police will have to deal more with LLMs being abused that way.
Bioweapons will be easier than in the past. LLMs will lower the barrier to entry, but bioweapons are hard to create and much of what the superpowers learned thankfully isn't in the training set for LLMs. I doubt there is much that can be inferred by having agents learn biology from first principles.
Ultimately, governments are responsible for policing these threats. Sadly we are currently both cutting government systems which work different aspects of these problems and withdrawing from the multinational orgs (UN, WHO, etc) who do lots of the investigating and watchdog work that underpin lots of the US's intelligence related to these fields.
The US has a schizophrenic regulation policy of AI where we both want to sell Nvidia chips to China (David Sachs) and we don't want to sell the top chips to China (bipartisan policy prior to Trump). We also have a terrible AI regulation policy where David Sachs disables models on a call-to-CEO-on-a-Friday-evening basis after Andy Jasse calls him with a scary story which turns out to be missing lots of relevant info (eg. The exploits was discovered in Mythos, not Fable, and other open weights models were able to find the same exploits).
There's not much we can do at a government policy level while Trump is snoozing through the rest of his term. So we are dependent on the AI companies to police themselves, but it's clear from this report that it's insufficient to prevent all serious abuse of the models.
> Simple enough to strip out anything that isn't a flag.
But it's not simple enough for every company/app to independently research the entire Unicode code point space (which is MASSIVE) to find out what kinds of "fix ups" that app needs to do to clean the data it consumes.
It's complicated or at least has difficult tradeoffs. Maybe you invest a lot of time carefully surveying all of the Unicode planes and decide which ones you care to keep unchanged and which ones you filter/strip. For every code point you reject or change, there is going to be some user who is confused or dissatisfied with the limitations of your app.
Maybe you intended to reply to my comment's parent? I don't necessarily disagree.
But also, according to the Ars article comment describing the Unicode character range, it has been deprecated, so maybe someone involved in the Unicode standard saw problems with it.
My suspicion is that the spam filter programmers didn't do a comprehensive evaluation of every code point on every plane of Unicode because... well that's a massive job. So your "sanitize and denormalize" tasks are actually massive mappings which were likely imperfectly created.
I'm basing my comment on the article. Since they are susceptible to ASCII smuggling, they can't have been doing unicode normalization.
The task isn't that massive. Python's unicodedata (for example) contains all the info already. Including the normalization (not denormalization as I wrote before) function.
> The Democrat Party believes everything that God hates. Like, we’re dealing with maggots, vermin, and swine here. This is not — these are not good people, they’re not even a little bit good.[1]
He also called for the death of Biden (because he was a "tyrant") and all surgeons of transgender people.
If he was merely trying to be offensive, then he succeeded.
If, however, he was trying to make someone fear enough for their life to preemptively defend himself from someone claiming to speak for God + using the same dehumanizing words that have preceded multiple genocides before, then he succeeded.
A "valid" response entirely depends on what you think he was apart of. If you think he was only involved in good faith debates on college campuses and getting out the vote, then obviously shooting the guy was unjustified. But he wasn't only that, so we have to struggle with what his legacy actually was and what might have driven someone to pick up a gun against him.
My point isn't to justify his killing. It's to point out that we need to learn from history so we don't blindly repeat it.
I don’t think the point of GP’s post was to claim one way or the other; rather it’s saying that it’s important to understand the motivations of others if you’re going to publicly characterize their actions or effectively oppose them.
This is a weirdly dissonant comment that misunderstands both religion and the nature of choices.
Religions don't exist without the people. The scriptures have no narrative power if they are not told by people to people. The rites and traditions only serve to reinforce habits and pass themselves down throughout time to new people. The religion evolves with the people because keeping followers become the fitness function.
No religion knowingly offers its followers "bad teachings". Anybody who thinks a religion has "bad teachings" quickly becomes an impostor and gets pushed out. The self selection means that there are no "bad teachings".
Also, you are trying to pretend like the choice is only applied by the cognitive brain by the individual human at the time of an action. This ignores both group psychology and the role of habits / indoctrination.
The Catholic Church internal policies target the ages of 4-14 for indoctrination and habit forming. What 10 year old will resist their parents, resist their church authority figures, resist all of the teachings of thousands of years of their people, resist the "word of God", have the knowledge and self awareness to know the difference between God's voice and typical bipolar ruminations?
> No religion knowingly offers its followers "bad teachings". Anybody who thinks a religion has "bad teachings" quickly becomes an impostor and gets pushed out. The self selection means that there are no "bad teachings".
I mean "bad teachings" from the point of view of an outsider who might not agree with the religion. What I'm saying is that even where that's the case, it's the people following the religion choosing to do so and choosing how they behave.
And I'm not pretending anything: I know full well about group psychology and indoctrination. But it still boils down to people being people, and people being hypocritical.
> The Catholic Church internal policies target the ages of 4-14 for indoctrination and habit forming.
Exactly. And what did I say about letting leaders off the hook? Grown adult church leaders choosing to indoctrinate, brainwash, and sometimes exploit children. It's people. Again. It's all people.
Sooner or later we all have to grow up and take responsibility and accountability for our own actions.
The current Trump Admin can't even decide on the first question, let alone come up for a plan on the second.
The Trump Admin's EO was to ask nicely all of the AI companies to give them 30 days to voluntarily review each model before wide release, but they have also failed to do that for the Mythos/Fable release, only to get a call from Amazon's CEO to David Sachs to convince them to disable Fable (to all non-US citizens).
We elected the party of "minimum regulations" into all 3 branches of government and we will reap what we sewed.
reply