EU AI Act To Target US Open Source Software

General Discussion
1

May 13, 2023
EU AI Act To Target US Open Source Software

In a bold stroke, the EU’s amended AI Act would ban American companies such as OpenAI, Amazon, Google, and IBM from providing API access to generative AI models. The amended act, voted out of committee on Thursday, would sanction American open-source developers and software distributors, such as GitHub, if unlicensed generative models became available in Europe. While the act includes open source exceptions for traditional machine learning models, it expressly forbids safe-harbor provisions for open source generative systems. While the act includes open source exceptions for traditional machine learning models, it expressly forbids safe-harbor provisions for open source generative systems.

Note that “traditional” means “non-generative” and for those of us who pay attention to the word games played in the ML hysteria, “generative” is ill-defined in such a way that it could exclude ANYTHING. It’s a buzzword without technical denotation – just as the entire ML industry is in a state of insolent hysterics that is exemplified by this guy in Yannic’s GPT4 Discord channel:

image
image1920×1065 197 KB

Any model made available in the EU, without first passing extensive, and expensive, licensing, would subject companies to massive fines of the greater of €20,000,000 or 4% of worldwide revenue. Opensource developers, and hosting services such as GitHub – as importers – would be liable for making unlicensed models available.

This extends to any LLM where:

…the output produced by the system is intended to be used in the Union…

…Open Source LLMs Not Exempt: Open source foundational models are not exempt from the act. The programmers and distributors of the software have legal liability…

… apparently even programmers that slap a “safe harbor” clause on their license.

10 Likes
2

In case this loathsome legislation requires any additional condemnation, the World Economic Forum thinks it’s just tickety-boo.

What’s going on here is that “European lawmakers”, having been blindsided by the Internet, which they failed to appreciate would bypass their control of media and enable the emergence of dread “populist” movements, are bent on keeping it from happening again through grass roots open source artificial intelligence discovering and disseminating “hate facts” from which their subjects must be protected.

7 Likes
3

The bottom line strategy appears to be to impose something like this on all uses of language models:

Language Model License Terms
blah blah boilerplate
You will not make language models that say hurtful things. Hurtful things are things that hurt as determined by a bunch of pubescent girls of any sex or age in some jurisdiction that can throw your ass in jail.
blah blah boilerplate

3 Likes