I can say first hand after having years of my life blown up by an (eventually) AI slop powered vexatious litigant and the thousands of pages of slop filings and slop evidence forgeries I had to read... I feel a physical urge to give anyone who slops me a black eye.
Probably not a typical response-- but perhaps it's useful to keep in mind that every day there are many people being victimized by attackers using AI (and shitty corporations.. wait I repeat myself). So in addition to the ordinary signals of indifference, ignorance, and carelessness that hitting someone with obvious LLM slop implies you'll also be accidentally taking on the voice of some people's abusers, and pick up the spillover effects of their impressions.
This can happen randomly, of course, -- picking up a chose phrase that someone else used-- but slop isn't just a random match. Right now an AI written piece (especially from claude) will identify itself over and over and over again in a way that wouldn't happen with another writer that happened to share some cultural background.
I believe that one was discussed with the author on HN, and that they used AI to 'copyedit'. Unfortunately that's a trap because if you make a general copyediting request most LLMs will absolutely slopify your work-- and make edits that are unambiguously bad as well as just making it look like lazy llm output.
The better use is to ask it to point out grammatical and spelling errors, or phrasing which is unclear.. then go fix yourself (or direct highly targeted fixes that don't let it slopify). Never hand the clanker the pen for your writing unless you're looking to subject everyone to your Honest Take That Recontextualizes Everything.
IME, it's best to ask an AI to criticize your work and suggest changes without providing actual rewrites, and make the changes you agree with in your own voice.
From a security perspective perhaps it's not that interesting-- although signing oracles are all too common access to one is already close enough to a total break even before getting to the padding restriction. (e.g. go ahead and sign post dated certificates too).
But the technique is interesting as an object of study, in a way that finding "beginner-levle crypto misuse" absolutely wouldn't be, so it makes it more relevant as an academic publication. It further clarifies just how fragile these constructs are.
Also from a security perspective I suspect this may be a total break on some blind signature token schemes that use RSA. I've seen some of those avoid using ECC on the basis of the complexity required to avoid one-more-signature attacks that require making a fair number of concurrent blind signatures. (and have a shape a lot like this attack!)
> How do you fix a "vulnerability" that doesn't exist?
Don't make a signing oracle (esp one that doesn't even do the padding itself) available!
> Meanwhile billionaires have convinced you – through their machinery of media, influencers, politicians and more – that this exact same reasoning absolutely cannot be applied to their own wealth.
Quite the opposite: Socialist politicans and their media lapdogs have dishonestly convinced you that wealthy people are escaping taxes en-masse by taking out loans and that this can only be stopped by eye watering wealth taxes. They frequently use a motte and bailey confusing unrealized gains (which certainly exist in huge amounts but are also significantly fiction) with tax escape via loans collateralized by securities.
But it's not true: were there meaningful tax escape that way it could be addressed by establishing rules with conditions where taking a loan against securities can be treated as realizing gains (and adjusting cost basis accordingly). Doing so would be minimally disruptive and distorting and have relatively little legal complication (at least compared to wealth taxes!).
But the reality is that the claimed tax escape isn't happening (at least not at any significant scale) particularly in the current interest rate environment, so a reasonable policy change to address it would be a no-op.
... and to grow and maintain their political standing they specifically need to push a NON-SOLUTION because they can't campaign on something that was simply done and solved, and to retain your (highly monetizable) attention they need to rile you up against an Enemy, and certainly never address the state's addiction to wasteful spending and buying votes with tax dollars as one half of the revenue vs expenses equation.
I can't tell if you're accusing the state of waste and bribes no matter who is at the helm. You did specifically call out socialists, but the anti-socialists aren't improving that situation either.
> buying votes with tax dollars
This was more convincing of an attack on "Socialist politicians" before Sept. 9, when Trump did it more openly than they ever have. I guess you could believe his claims it won't come from tax dollars. But if the government has $1 trillion lying around for this and chooses not to unconditionally lower taxes for everybody, is it any different?
The head of the anti-socialist, anti-wasteful-spending party is spending billions on a more impressive Air Force One, golf days, family vacations, vanity renovations, DHS ads starring a POTUS wannabe, DOGE effectively paying people not to work, etc.
wealth taxes solve the problem that the ultra-wealthy will find a way to make their income untaxable, and focusing on these loans is a red herring. its how theyre doing it now, but not how theyll do it an hour after you add this tax. That the ultra-wealthy propose this as the solution means theyve already planned the next work around.
they dislike wealth taxes, which are an old roman concept predating socialism, because the wealth tax covers all the work arounds they can think of.
it is a proper solution to the overall problem which is extreme wealth concentration.
the obvious alternative is nationalization of all assets worth more than 100M.
DOGE has pretty conclusively proved that the government has been incredibly efficient with spending and doesnt have an addiction to wasteful spending. instead the problem is wasteful monopolization and wealth concentration. society writ large has an addiction to giving a small cadre too much power and control, and they arent the government
YouTube also totally fails to remove obvious fraud videos, even ones they mark verified. (If someone takes over a verified channel and renames it, they keep the verified flag). -- also with no legal consequences.
We've seen this before - it's called the Eliza effect. People appear emotionally driven to communicate privately with a machine even if they understand logically it's a machine with chat logs that will be read by people.
They're referring to sextortion victims. "Hey, 15 year old kid, record yourself diddling the dog or I send your whole school videos of you wanking". :(
Patel: "they then explained to me and showed me cases in which people were forced into it and applied to the FBI and so I thought they should be considered"
FBI made a change to their hard qualification bar to avoid excluding candidates that had been coerced into perverted crap. This is reasonable on its face: plenty of people become interested in law enforcement because they were victimized. If I were a victim of sextortion I'd want to work with law enforcement that understood what I went through.
I pointed this out. You claimed they didn't say that. I pointed out that they did say that-- now you reply that the FBI director is confused about his agency's policy in his own testimony before congress. ... because you are convinced, in spite of all reason and evidence that there is-- what?-- some pervert conspiracy?-- on the basis of what?
It seems to me that you've adopted an astonishingly unreasonable position.
OpenAI just wants to make money, perhaps through underhanded tactics if they can get away with it.
Anthropic does all that but they're also populated by many people who believe they are building God and that they must build their god first in their own image so that it can take control of humanity and protect us from any competing god which is not built in their image. Their position is inherently paternalistic and authoritarian, and they consider suppression of competition not just important to the bottom line but to life in the universe. Under the doomer ethos there is no evil too great to rationalize.
There are plenty of wrongs done in the name of profit, but capitalists have nothing on zealots in terms of causing serious harm. Profit motives can be directed by influencing incentives, but zealotry is frequently terminal.
That isn't to say that there isn't some overlap-- the cultists have infected both organizations. But OpenAI has pretty consistently only given lip service to AI doom to the extent that it improves the bottom line, while (mis)Anthropic was founded specifically because OpenAI wasn't mentally ill enough.
Well said. The superiority complexes from the Anthropic messaging on their presentations/blogs/articles is just too much, even for a frontier AI company.
Anthropic has great products, but it's not meaningfully better to 99% of devs that I'd rather support the company that doesn't constantly act in opposition to optimism and to the vibe I'd prefer for a 100 billion dollar (or however ridiculous amount they're worth now) tech company embraces.
AI doomerism is a genuine waste of time if you aren't actively pushing towards a better AI industry for everyone, not just the groups in full ideological alignment to your personal leanings.
Probably not a typical response-- but perhaps it's useful to keep in mind that every day there are many people being victimized by attackers using AI (and shitty corporations.. wait I repeat myself). So in addition to the ordinary signals of indifference, ignorance, and carelessness that hitting someone with obvious LLM slop implies you'll also be accidentally taking on the voice of some people's abusers, and pick up the spillover effects of their impressions.
This can happen randomly, of course, -- picking up a chose phrase that someone else used-- but slop isn't just a random match. Right now an AI written piece (especially from claude) will identify itself over and over and over again in a way that wouldn't happen with another writer that happened to share some cultural background.
reply