In short, we believes LLMs are going to be a part of the future of int'l lawyering whether we want them to or not ... it's now up to the profession to work through & test how well they can do so, to embrace their potential but remain clear-eyed about the risks LLMs pose.

Finally, we worry about LLMs being 4) corruptors - i.e., given automation bias, they may become "the" source for int'l law whether their answers would be the same as analog int'l lawyers might offer. & if that's true, some may look to poison data sets or tilt the scales to get their desired answers

We also see LLMs as 3) creators, legitimately and creatively offering first drafts or ideas for int'l lawyers (especially those just getting up to speed on a topic)

In the end, we see 4 possibilities for LLMs - 1) as a collaborator that can quickly afford first takes and summaries for int'l lawyers; but also 2) as a confounder with its well-known propensity to hallucinate sources or send lawyers down rabbit holes.

Specifically, we examine how well LLMs do identifying custom and/or creating treaty texts (& arguments to persuade specific states to join them). We also explore how LLMs might be employed to interpret or apply int'l law & to digest & summarize large datasets (e.g., all those amici briefs @ the ICJ)

The caveats are that any threat to use force would need to involve force that would be unlawful (a box pretty easily checked) AND credible (trickier but hard to say it shouldn't be presumptively treated as such given the office of the source). #LawSky

Indeed, combined with the AU's path-breaking common position, we've certainly come a long way from Harold Koh's 2012 speech on the USG position on how IL applies in cyberspace

Overall, I think the statement is a very welcome development & EU Member States deserve praise for getting at least some form of a common position out into the public sphere.