I posted this on HN yesterday, with a comment saying I was skeptical. Then we talked about it on Slack and Thomas Pornin utterly debunked it, so I deleted it. Pornin posted his analysis this morning:

https://crypto.stackexchange.com/questions/70467/what-are-th...

The cryptanalysis (really, "proof of zero knowledge", if I might clarify and shoplift someone else's zinger at the same time) they claim to have performed is slower than brute force.

Long story short: this is almost certainly just a troll.