Where this was said
AI Breaks the Deadlock: Vitaly Decodes FAST16
At 14:34 · chapter starts 12:22
Fast forward to the present day: Jags runs a large research team and wonders whether new AI tools could solve what years of human effort couldn't. He assigns the FAST16 problem to Vitaly Kamluk, a Belarusian cybersecurity legend based in Singapore, known for his calm, Zen-like demeanor. [1] — Juan Andrés Guerrero Saade "Jags's colleague Vitaly Kamluk disappeared for two weeks into the FAST16 problem, using AI to verify his reverse engineering. What emerged:…" 12:20 Rather than jump straight to testing the AI, Vitaly first spends nearly two weeks doing the reverse engineering himself — going completely dark, not answering messages — so he'd know whether the AI was getting it right. Then, at around 1am Singapore time, Jags gets a message: 'We need to talk.' Vitaly tells him he's had the AI models double and triple-check his work, and they all agree. The verdict: FAST16 targeted floating-point, high-precision math calculations — a type of operation no malware had ever touched before, and one exclusively associated with complex scientific modeling. Hearing this from the usually unflappable Vitaly, Jags takes it very seriously.
Vitaly Kamluk spent approximately two weeks in near-total isolation reverse-engineering FAST16, then used AI models to double and triple-check his findings.
FAST16 was developed around 2005, the same era as Stuxnet, and shares similar architecture though no shared code.
FAST16 is the first piece of malware Jags had ever encountered that specifically targeted floating-point, high-precision mathematical calculations.
