Thank you, you're correct. Effort.news was one of my research sources, but you're right that although OpenAI use Irregular, they were not involved in the specific HF incident (although the failure mode was otherwise identical). I've updated the post to make that clear.
replyAs of writing it still says:
reply> For Anthropic, Google, and Meta, the catastrophic breakouts happened inside the testing environments of the exact same contractor.
If this is the level of understanding you have of the relevant incidents, there's a lot of chutzpah in saying that other people are "selling garbage", carrying out an "extraordinary confidence trick", etc.
[dead]
replyThe failure mode was _not_ identical. The HF incident agents were not directly connected to the internet and had to compromise an internal package registry in order to access the internet.
replythank you for this reasonable reaction
replyCan you point out an inaccuracy in the effort.news piece on the hacking incidents? HuggingFace only appears once, as a "similar", not levied against Irregular
replygenuinely curious, haven't heard others raise any yet, but does not mean it is issue free