←back to thread

989 points acomjean | 1 comments | | HN request time: 0.289s | source
Show context
aeon_ai ◴[] No.45143392[source]
To be very clear on this point - this is not related to model training.

It’s important in the fair use assessment to understand that the training itself is fair use, but the pirating of the books is the issue at hand here, and is what Anthropic “whoopsied” into in acquiring the training data.

Buying used copies of books, scanning them, and training on it is fine.

Rainbows End was prescient in many ways.

replies(36): >>45143460 #>>45143461 #>>45143507 #>>45143513 #>>45143567 #>>45143731 #>>45143840 #>>45143861 #>>45144037 #>>45144244 #>>45144321 #>>45144837 #>>45144843 #>>45144845 #>>45144903 #>>45144951 #>>45145884 #>>45145907 #>>45146038 #>>45146135 #>>45146167 #>>45146218 #>>45146268 #>>45146425 #>>45146773 #>>45146935 #>>45147139 #>>45147257 #>>45147558 #>>45147682 #>>45148227 #>>45150324 #>>45150567 #>>45151562 #>>45151934 #>>45153210 #
rchaud ◴[] No.45144837[source]
> Buying used copies of books, scanning them, and training on it is fine.

But nobody was ever going to that, not when there are billions in VC dollars at stake for whoever moves fastest. Everybody will simply risk the fine, which tends to not be anywhere close to enough to have a deterrent effect in the future.

That is like saying Uber would have not had any problems if they just entered into a licensing contract with taxi medallion holders. It was faster to just put unlicensed taxis on the streets and use investor money to pay fines and lobby for favorable legislation. In the same way, it was faster for Anthropic to load up their models with un-DRM'd PDFs and ePUBs from wherever instead of licensing them publisher by publisher.

replies(15): >>45144965 #>>45145196 #>>45145216 #>>45145270 #>>45145297 #>>45145300 #>>45145388 #>>45146392 #>>45146407 #>>45146846 #>>45147108 #>>45147461 #>>45148242 #>>45152291 #>>45152841 #
ReFruity ◴[] No.45144965[source]
> But nobody was ever going to that

If this is a choice between risking to pay 1.5 billion or just paying 15 mil safely, they might.

replies(2): >>45145247 #>>45145248 #
crote ◴[] No.45145247[source]
Option 1: $183B valuation, $1.5B settlement.

Option 2: near-$0 valuation, $15M purchasing cost.

To an investor, that just looks like a pretty good deal, I reckon. It's just the cost of doing business - which in my opionion is exactly what is wrong with practices like these.

replies(2): >>45145553 #>>45146344 #
Joel_Mckay ◴[] No.45146344[source]
In most places, a legal settlement is considered a tax deductible loss. At a certain scale it will likely cost the company nothing, but these kinds of cases often trigger speculators grabbing discount stock from panicking amateurs. lol We still have no idea what they sell, so avoided exposure to their antics... =3
replies(1): >>45151914 #
tpmoney ◴[] No.45151914[source]
But that isn't how tax deductions work. Since taxes are always a fraction of income, a deduction can never save you more money than you already paid out to get the deduction in the first place. If you have a 10% tax rate, your options are:

A) Make 100M, pay 10M in taxes

or

B) Make 100M, pay 10M in lawsuit settlements, pay 9M in taxes

You come out ahead every time by not paying the settlement in the first place.

replies(1): >>45154099 #
Joel_Mckay ◴[] No.45154099[source]
You may be confused, but a business loss deduction usually reduces a taxable income. In general, most systems only require the cost/loss was incurred during gaining or producing income from a business or property.

For a significant sum, we should assume their team consulted a specialist firm on the subject at their location. People don't often YOLO this stuff at that scale, and businesses don't always settle every time they get shaken down for cash... some go to war, as it can be cheaper to sandbag/delay till opponents go bankrupt.

Have a great day. =3

replies(1): >>45155367 #
1. tpmoney ◴[] No.45155367[source]
> You may be confused, but a business loss deduction usually reduces a taxable income.

Yes, I know, which is why in option B the taxes required was $9M instead of $10M. The 10k payment reduced taxable income from $100M to $90M. Business taxes are notoriously complex, but I am aware of no IRS rules that would allow a 10M legal settlement to reduce the taxes owed by >= $10M. If you believe I remain confused, please by all means provide an example scenario and/or citations to the relevant tax statutes.