text stringlengths 0 2.35k |
|---|
[2867.60 --> 2868.88] On the other end, |
[2869.12 --> 2870.52] when you have |
[2870.52 --> 2871.64] some assembly code, |
[2872.00 --> 2872.82] then assembly code |
[2872.82 --> 2873.86] is going to be |
[2873.86 --> 2875.12] this very strict |
[2875.12 --> 2876.08] unique language |
[2876.08 --> 2877.32] that probably all AIs |
[2877.32 --> 2878.26] will have to work on |
[2878.26 --> 2879.74] and I'm not exactly sure |
[2879.74 --> 2880.70] how they are going |
[2880.70 --> 2881.66] to work their way |
[2881.66 --> 2883.20] back up to either |
[2883.20 --> 2884.32] recognizing a function |
[2884.32 --> 2885.76] or actually generating |
[2885.76 --> 2887.52] corresponding high-level code. |
[2888.38 --> 2889.20] But the good thing |
[2889.20 --> 2890.00] that AIs will have |
[2890.00 --> 2890.58] going for them |
[2890.58 --> 2891.32] is that assembly |
[2891.32 --> 2892.26] is going to be |
[2892.26 --> 2893.38] the exact opposite |
[2893.38 --> 2894.30] of ambiguous. |
[2894.78 --> 2896.20] You have ambiguous |
[2896.20 --> 2897.20] and at the exact |
[2897.20 --> 2897.90] other end |
[2897.90 --> 2899.46] of what possible language |
[2899.46 --> 2899.72] you have, |
[2899.74 --> 2900.28] you have assembly |
[2900.28 --> 2902.32] which is 100% precise. |
[2902.56 --> 2903.84] As consistent as it gets. |
[2903.94 --> 2904.56] Yeah, consistent |
[2904.56 --> 2905.86] and actually dumb |
[2905.86 --> 2906.56] in some ways |
[2906.56 --> 2907.64] but it's just |
[2907.64 --> 2909.30] very simple operations |
[2909.30 --> 2910.26] that can only do |
[2910.26 --> 2910.92] a single thing |
[2910.92 --> 2911.82] in a very defined way. |
[2911.96 --> 2913.54] So on that front, |
[2913.66 --> 2914.40] I think that this is |
[2914.40 --> 2915.02] actually going to be |
[2915.02 --> 2916.12] a very, very good thing |
[2916.12 --> 2916.80] for the AIs |
[2916.80 --> 2917.92] whenever they are ready. |
[2918.18 --> 2919.24] So there's already |
[2919.24 --> 2920.38] good tools out there |
[2920.38 --> 2921.28] that just take binary |
[2921.28 --> 2922.32] and translate that |
[2922.32 --> 2923.84] into assembly, |
[2924.10 --> 2924.88] not 100%, |
[2924.88 --> 2926.52] but a very good coverage. |
[2927.18 --> 2927.66] And assembly |
[2927.66 --> 2928.68] is consistent enough |
[2928.68 --> 2929.62] so that means that |
[2929.62 --> 2931.64] some IDAPro plugin |
[2931.64 --> 2933.26] that uses AI |
[2933.26 --> 2934.14] must be developed |
[2934.14 --> 2934.80] as we speak |
[2934.80 --> 2936.06] to say, |
[2936.30 --> 2936.48] like, |
[2936.52 --> 2937.56] here's an assembly input, |
[2937.56 --> 2938.58] but please translate that |
[2938.58 --> 2939.70] to GoCode for me. |
[2940.12 --> 2941.18] Well, it's a good question |
[2941.18 --> 2943.10] because you would think |
[2943.10 --> 2943.56] that someone |
[2943.56 --> 2944.64] would be working on this |
[2944.64 --> 2945.68] but when you look |
[2945.68 --> 2946.54] at the market |
[2946.54 --> 2947.88] for reverse engineering tool, |
[2947.94 --> 2948.84] it's actually quite small. |
[2949.02 --> 2950.50] You have Hexrays, |
[2950.70 --> 2951.96] the creators of IDAPro |
[2951.96 --> 2953.74] and Hexrays |
[2953.74 --> 2954.70] are kind of a |
[2954.70 --> 2956.00] old school company. |
[2956.20 --> 2956.24] Like, |
[2956.26 --> 2958.20] their product is amazing |
[2958.20 --> 2959.12] but, |
[2959.28 --> 2959.54] like, |
[2959.62 --> 2960.06] they've been, |
[2960.66 --> 2961.82] they haven't really tried |
[2961.82 --> 2963.24] to create |
[2963.24 --> 2964.34] any form of disruption |
[2964.34 --> 2965.62] in the past 20 years. |
[2965.80 --> 2965.94] Now, |
[2966.00 --> 2966.54] they have been doing |
[2966.54 --> 2967.24] lots of improvements |
[2967.24 --> 2967.88] to their products |
[2967.88 --> 2969.00] but I think part of it |
[2969.00 --> 2969.98] is only because |
[2969.98 --> 2970.92] they have been challenged |
[2970.92 --> 2971.76] by Ghidra, |
[2971.96 --> 2973.12] their open source competitor. |
[2973.78 --> 2973.90] Right? |
[2974.24 --> 2975.32] It's kind of |
[2975.32 --> 2976.82] my opinion that, |
[2976.98 --> 2977.12] well, |
[2977.14 --> 2977.68] let's make this |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.