4 ms·
Well flagging as “potentially malicious” seems fine and super useful. Companies just need to have competent investigation of the reports and avoid the dark side
by jasonthorsness 2y ago
Well flagging as “potentially malicious” seems fine and super useful. Companies just need to have competent investigation of the reports and avoid the dark side of automating action just because it’s cheaper.
- cratermoon 2y ago"just".[1] AI-powered bots can generate a flood of reports of "potentially malicious" code. No company could possibly keep up. Effective DDOS attack on security. Not even the NIST can keep up. https://www.theregister.com/2024/10/02/cve_pileup_nvd_missed_deadline/ https://www.theregister.com/2024/10/02/cve_pileup_nvd_missed... 1 https://sgringwe.com/2019/10/10/Please-just-stop-saying-just.html https://sgringwe.com/2019/10/10/Please-just-stop-saying-just...
- jasonthorsness 2y agoThis might be another problem waiting for a better model like most of the rest of the AI use cases today
- cratermoon 2y agoYeah and the next step to getting to the moon is building a taller ladder. https://thebullshitmachines.com/lesson-16-the-first-step-fallacy/index.html https://thebullshitmachines.com/lesson-16-the-first-step-fal...