Um, whoops there. Not true. Our lab did it before. Here's the video in the supplemental info from the 2006 paper. Granted that Andy's robot is more impressive.
Also, notice that one of the Nature videos for the monkey-bot paper covers the arm, while the others show the arm movement. Compare this to this. They're both clearly shot for mass consumption, because view of the headstage is obviously being obscured. If it's not a big deal, why hide it?
And finally, the supplemental info reports everything in much more detail. Like that the two monkeys had completely different electrodes (one array, one microwire), so that the number of cells used was significantly different (>100 vs <30), and it took a month of training. The microwire monkey had slightly worse performance (78% vs 61%).
There was also an NPR interview where he makes a couple false statements, and mentions doing stimulation of somatosensory cortex in future human implants, even though he has been highly critical of any use of the technology in humans. I don't understand the 'tude, dude.