| ▲ | wat10000 3 hours ago | |
I'd be surprised if that's what happened. The instant I see "corrupting the heap" I assume it's a serious vulnerability unless proven otherwise. AI is taking it to another level, but the days of dismissing memory corruption as unexploitable died years before. | ||
| ▲ | eru an hour ago | parent | next [-] | |
Yes, even before LLMs it was a common heuristic that if you can make something crash, with a bit more cleverness you can usually take over. | ||
| ▲ | cryptonector 2 hours ago | parent | prev [-] | |
100% this. | ||