o fim dos jogos — Transcript
Full transcript
- 0:06A huge debate recently erupted on the
- 0:09internet because of a mod that put
- 0:12Spider-Man into a Batman game in record
- 0:16time. Even though, as always, there was
- 0:20a lot of fake news circulating, this
- 0:23situation is bizarre on a surreal level
- 0:25, as it involves a technology that is
- 0:28panicking the cybersecurity industry:
- 0:30AI-driven reverse engineering. So,
- 0:36through it, people are putting one game
- 0:37inside another. Literally, first-person
- 0:42shooters, like COD, and skateboarding
- 0:45games have completely opposite physics
- 0:47engines, but modders used AI agents to
- 0:50extract the board physics, ramp physics
- 0:53calculations, and the trick and gravity
- 0:56systems from the skate game and inject
- 0:59them directly into the COD map. The
- 1:04character now rides a skateboard and
- 1:06performs maneuvers down the stairs of
- 1:07shooter maps with the physics of the
- 1:09skating game. It's bizarre. The Elden
- 1:13Ring engine is made for hyper-realistic
- 1:15graphics and complex sword combat.
- 1:18Minecraft, on the other hand, runs on
- 1:20cube graphics with procedural world
- 1:22destruction and construction. The AI
- 1:25helped map how Elden Ring reads ground
- 1:28collision and translated the Minecraft
- 1:30system. So, the player can mine the
- 1:34Elden Ring landscape, grab blocks,
- 1:36store them in their inventory,
- 1:38Minecraft-style, and build bridges to
- 1:40escape bosses. It’s eerie. Batman
- 1:43doesn't shoot webs or swing using
- 1:45Spider-Man's physics pendulum. But the
- 1:48mod doesn't just change Batman's look.
- 1:51The AI mapped how the Spider-Man game
- 1:53calculates web tension on a building's
- 1:56roof, the arc angle, and momentum,
- 1:58allowing those calculations to be
- 2:00injected directly into the Arkham
- 2:02Knight map. It’s eerie,
- 2:05>> so. But I really trip out on this, man.
- 2:07>> To understand the scale of this, think
- 2:09of any game or app as a finished cake.
- 2:13When a company releases software, they
- 2:15give you the cake, but hide the
- 2:16original recipe, the so-called source
- 2:18code. They do this to prevent copying,
- 2:22piracy, and cheating. To hide this
- 2:25recipe, in a very basic way, the
- 2:27company puts the source code through a
- 2:29process called compilation. When
- 2:32programmers create a game, they write
- 2:35everything in human-readable languages,
- 2:38like C ++ or Python. But the computer
- 2:41doesn't understand human words.
- 2:44Therefore, a compiler program takes all
- 2:46that text and chews it up, converting
- 2:48the game logic into machine language.
- 2:52An infinite mountain of zeros and ones,
- 2:55which are binary codes, ultra-raw
- 2:57instructions called assembly. By doing
- 3:01this, the compiler intentionally
- 3:03deletes everything that would help a
- 3:05human understand the original structure
- 3:07. it erases the names of variables and
- 3:10functions. So, a function that in the
- 3:13original recipe was named something
- 3:16like, for example, calculate shot
- 3:18damage, becomes just something generic
- 3:21and unreadable, like sub_line 40100.
- 3:25Then, it also removes comments and
- 3:27notes. So, all the explanations the
- 3:31devs left to organize the code are
- 3:33destroyed, and finally, it scrambles
- 3:35and optimizes the logic. The compiler
- 3:39reorganizes the instructions to make
- 3:41the game run as fast as possible on the
- 3:43processor, which makes the internal
- 3:45structure even more confusing and messy
- 3:48. Beyond basic compilation, many
- 3:51companies apply extra layers of
- 3:53protection called code obfuscation and
- 3:56anti-piracy systems like DRM or Denuvo.
- 4:01They place intentional knots in the
- 4:04code, create fake routines, and encrypt
- 4:07parts of the exe file so that even if
- 4:09someone tries to open the game from the
- 4:12inside, they only find a jumble of
- 4:15nonsensical data. And that is exactly
- 4:19why turning the game back into the
- 4:21original recipe, which is the process
- 4:23of reverse engineering, has always been
- 4:25one of the most difficult and
- 4:27time-consuming tasks in computing.
- 4:29>> It takes time. And that is where the
- 4:31danger lies. Artificial intelligence
- 4:35undoes all this scrambling, acting as
- 4:38an instant translator and a pattern
- 4:40detective. If the compiler took the
- 4:45original text, destroyed the names, hid
- 4:48the logic, and turned everything into a
- 4:50heap of undecipherable zeros and ones,
- 4:53the AI reverses the path through three
- 4:56main steps. The first would be pattern
- 5:00recognition, the code's signature. The
- 5:03AI doesn't try to guess what the code
- 5:05does from scratch. It was trained by
- 5:07analyzing billions of lines of code
- 5:09before and after they were compiled.
- 5:13When it looks at that tangle of generic
- 5:15instructions, like sub_line 40100, it
- 5:18recognizes the mathematical structure.
- 5:22It identifies the way these numbers
- 5:24move in memory. 99%identical to the
- 5:27equation that calculates the gravity of
- 5:29a falling object, for example. Yeah,
- 5:32many people still don't have faith in
- 5:34artificial intelligence. AI, man, is
- 5:36the Antichrist himself, you have no
- 5:38idea. I'm still going to make a video
- 5:40about this. This stuff is getting
- 5:41bizarre. The second step it uses is
- 5:44deobfuscation and automatic naming,
- 5:46which would be a translation, right,
- 5:49into human language. The most
- 5:51time-consuming step for a human
- 5:53programmer was figuring out what each
- 5:55variable meant. And the AI solves that
- 5:57in seconds. It reads the routine and
- 6:01returns the logical names, renaming the
- 6:03ghost variable. For example, variable
- 6:07underscore 1a 2b. It will translate it
- 6:10to player speed and the function
- 6:12sub_line 40100 to calculate web inertia
- 6:15, things like that, you know? Then it
- 6:19rewrites the raw assembly instructions
- 6:22back into a clean, organized language
- 6:24like C ++ or Python, even adding
- 6:26explanatory comments for the mod. The
- 6:31third step is real-time memory mapping.
- 6:35So, besides translating static code, AI
- 6:38tools connected to the game while it
- 6:41runs can monitor the RAM. If a modder
- 6:45wants to inject Spider-Man's web into
- 6:47Arkham Knight, it locates exactly the
- 6:49memory address where Batman's physics
- 6:51processes position changes. Next, it
- 6:55generates the bridge script, which is
- 6:57the code injection that forces the
- 6:59engine to read and execute the
- 7:01Spider-Man game's movement equations as
- 7:03if they were native to the game itself.
- 7:06And man, this isn't just dangerous
- 7:08because the AI broke through the game's
- 7:10protection barriers. It's dangerous
- 7:12because it broke the barrier of time
- 7:14and technical expertise, creating a
- 7:16systemic risk for the entire digital
- 7:18infrastructure. The real danger isn't
- 7:21having Spider-Man in Gotham. The danger
- 7:23is the absurdly fast speed of the
- 7:26automation. In the past, to
- 7:29reverse-engineer complex software, like
- 7:32a banking system or a AAA game, you
- 7:34needed to be a computing wizard, a
- 7:37senior expert who spent decades
- 7:39studying languages and memory analysis.
- 7:43There were few in the world capable of
- 7:45this, but in the future, right? I hope
- 7:48not, but it could happen. Anyone with
- 7:51basic programming knowledge and access
- 7:53to these AI tools can cause the same
- 7:56damage. AI has become a force
- 7:59multiplier for malicious minds. If we
- 8:03previously had at most 100 people in
- 8:05the world capable of breaking complex
- 8:08protection, now we potentially have
- 8:10millions, and that is extremely
- 8:12dangerous. Most of the world's software
- 8:16, including mobile apps, corporate
- 8:18programs, and industrial systems, bases
- 8:20part of its security on the fact that
- 8:23its code is closed, compiled, and
- 8:25obfuscated—scrambled, as I mentioned.
- 8:28Companies rely on the fact that it
- 8:30would take too long for someone to
- 8:32disassemble the program and find a flaw
- 8:35. AI has eliminated the time factor. An
- 8:39AI agent can analyze an entire
- 8:41executable in search of zero-day
- 8:43vulnerabilities. These are flaws that
- 8:46nobody knows about yet, doing in hours
- 8:49what would take a human months or even
- 8:51years. If an AI can find a game's
- 8:54physics function in 8 hours to create a
- 8:56mod. It can find an app's password
- 8:59validation function to bypass it in the
- 9:01same amount of time. Of course, this
- 9:03isn't happening yet, but in the future,
- 9:05measures will have to be taken because
- 9:07it's bizarre. Security systems, such as
- 9:10anti-cheats in online games or
- 9:12antivirus programs, function largely by
- 9:15recognizing signatures. They know
- 9:18exactly what a known hack or virus
- 9:19looks like in the code. AI can be used
- 9:22not only to create the hack but to
- 9:24rewrite it automatically every 10
- 9:26minutes. It can generate thousands of
- 9:29variations of the same malicious code,
- 9:31changing its digital signature so
- 9:34defense systems can never identify it.
- 9:36This creates a cat-and-mouse scenario,
- 9:39where current defenses are too slow to
- 9:41keep up with the speed of AI-generated
- 9:43attacks. Community projects that once
- 9:47took 10 years to release, like a RE4
- 9:49Remake or, I don't know, a Resident
- 9:51Evil 2 Remake, could be done by small
- 9:53teams in a few months. Forgotten PS1 or
- 9:57PS2 games that don't run on modern
- 10:00computers could have their code
- 10:02decompiled by AI and be ported natively
- 10:05to PC with 4K, 60 FPS, and ray tracing
- 10:08support in record time. The player will
- 10:11no longer depend on the whims of the
- 10:13developers. Want to play Minecraft
- 10:16inside GTA V? AI will make that
- 10:18accessible. Of course, big companies
- 10:22like Sony, Nintendo, and Rockstar won't
- 10:24just stand by and watch this happen.
- 10:27What happens to the legal issue of
- 10:29using Sony's proprietary Spider-Man
- 10:31mechanics running inside a Warner game?
- 10:34So, it is very likely, almost certain
- 10:36actually, that we will see updates to
- 10:39terms of service and new legal
- 10:41proceedings against tool creators and
- 10:43AI decompilation. To prevent AI from
- 10:46reading and modifying the executable. X
- 10:49on your computer. Developers may
- 10:52accelerate the migration to games where
- 10:55all logic runs in the cloud, killing
- 10:57traditional single-player game modding.
- 11:02Companies are already using AI models
- 11:04to create dynamic encryption, right?
- 11:06Meaning the game code changes its form
- 11:09in memory while the game is running. So
- 11:13that the attacker's AI cannot find the
- 11:15patterns. AI in reverse engineering has
- 11:18therefore opened this Pandora's box. On
- 11:21one hand, we have a golden age of mods
- 11:23and limitless creativity for players.
- 11:27On the other, a real risk to the
- 11:28security of online games and our
- 11:30everyday software. Do you think
- 11:33developers should ban this type of tool
- 11:35? Or is the advancement going to be
- 11:38inevitable? Comment down below, because
- 11:40I want to know. Don't
About this transcript
This page contains the full transcript of o fim dos jogos by JJ, generated from the public captions YouTube serves with the video. The transcript has 1,625 words across 263 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.