Anthropic, which has a 'proven track record' of every once in a while advertising with fear, saying how dangerous their new product is in one way or another (a year ago it was able to blackmail users, after they prompted it like 200 times to get their desired result), but what Anthropic's new model is doing is coding better, reading code better, and finding vulnerabilities that no one has seen for years of reviewing code, and it can chain multiple vulnerabilities to maybe create an exploit.
One possibility not mentioned on the show is that these are the exploits that the government agencies had in pocket, and now they might close them up.
How long before people's PCs are just offline, no connection possible, machines?