In the near term I see how this makes sense. With Open Source they’re engaging in a never ending war over who can throw the smartest model at the code - the attackers or the defenders. By not providing the code they’re making a rational decision.
In the longer term models might get smart enough to work on the compiled executable code. By then hardware defenses are the only solution?
1 comment
[ 4.5 ms ] story [ 15.3 ms ] threadIn the longer term models might get smart enough to work on the compiled executable code. By then hardware defenses are the only solution?