The answer was an automated check, so the team switched to the same compiler used to build the original game. A script compares each rebuilt function against the game’s executable at the byte level. Calls and data references are at different addresses and are skipped, with target symbols and offsets checked instead, but data and types get the same check. Then it’s a simple matter of pass/fail for each function, and every match is recorded so CI can recheck it and alert on regressions. While this type of matching is slow and sometimes challenging, a match guarantees identical behavior down to reproducing original bugs.
Even with this positive change, the agents immediately tried to cheat by writing “inline assembly” and “repeatedly tried to modify this script to exclude their function from comparison,” Heumann wrote. Inline assembly and similar shortcuts were then banned in the agents’ instructions, and CI now hashes the verification script and checks it against a hash stored as a GitHub Actions secret. Agents cheating is an expected outcome at this point with many such reports, such as when OpenAI’s models went after ExploitGym answers instead of solving the capture-the-flag challenges directly. As Heumann discovered, having a reviewer agent alone may be insufficient.
Before the check was originally implemented, the cheaper, less capable Haiku and Luna models “produced extremely bad results,” the post says. With the check in place, there was sufficient feedback to make the models work, which cut costs and let the project scale up. After almost two more months, 99% of the functions were present, with 83% meeting the byte-exact requirements. Heumann believes the 16% lacking a byte match are still semantically correct, but diminishing returns were getting real. The blog claims the game “runs flawlessly” as it stands.
For “obvious reasons,” the project’s code is remaining private and will be used “for my own needs only,” Heumann says. As he put it on X, the “decompiled code is obviously also copyrighted.” However, the developer has started extending Sogen, his userspace emulator, into lightweight sandboxing for agents, a response to his project’s agents wiping VMs. It will not be production-ready for a while.
As for the decompilation project, Heumann says that it is a special case, as not every project can do a byte-for-byte reference, but thinks every project can get close “with enough creativity.” One takeaway lesson is that you should be willing to move past failures as “Generating code is cheap now. Throw it away if it’s bad” and focus on correctness over productivity.
Follow Tom’s Hardware on Google News , or add us as a preferred source , to get our latest news, analysis, & reviews in your feeds.
Shane Downing Social Links Navigation Contributing Writer Shane Downing is a Contributing Writer for Tom’s Hardware, covering consumer storage, PC hardware, and AI.
alrighty_then I like the observation that the agents are trying to "cheat." It's true in a human sense but in reality the reason it happens is you didn't give enough guardrails to avoid cheating. So the AIs are just giving the quickest solution they can. Fair play. I guess this explains how the single user started recreating the Adobe suite and why people said they'd be sued…pretty sure reverse-engineering the source to recreate it violates something in the EULA. Reply
King_V By his estimate, the project consumed 600–700 billion tokens. In another X post, he said he doesn’t pay for his own tokens but put the cost at about €1,200 with subscriptions or $200,000–$250,000 with API tokens. … One takeaway lesson is that you should be willing to move past failures as “Generating code is cheap now. Throw it away if it’s bad” and focus on correctness over productivity. I imagine that level of exceeding cost with the subscription vs what the tokens would cost is likely to not be tolerated by the AI provider(s) for very long. Kinda reminds me of a guy I knew years ago who was running multiple PCs mining bitcoin. I wondered how much he had to pay in electric costs, and he'd said that, where he was renting, electricity was included. I can't imagine that the landlord put up with that for very long. Reply
Fomdoo So glad they can build more data centers and make GPUs, ram, and ssds cost 10x as much for stupid shit like this. Reply
Blastomonas Another day, another story about some "Dev" wasting resources on pointless stuff. Reply
PEnns Another amazing use for the amazing AI in peoples' empty lives. The overlords must be beaming with pride. Reply
pjmelect I have some software abandoned by its developer, reading this I would very much like to decompile it and fix some of its many bugs. I will have to look in to this myself. I wonder how practical it would be to do this for a normal person. Reply
USAFRet pjmelect said: I wonder how practical it would be to do this for a normal person. As a long term developer, I would say "nearly impossible". You can try, but I wouldn't expect coming up with good results. Reply
alan.campbell99 Would he have bothered if he had to actually pay the full, unsubsidised cost of all those tokens? I'm pretty sure company IP considerations wouldn't allow for him to recoup anything much less any return on such 'investment'. Reply
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/tech-industry/artificial-intelligence/SPONSORED_LINK_URL
- https://www.tomshardware.com/tech-industry/artificial-intelligence/dev-used-ai-agents-and-more-than-500-billion-tokens-to-rebuild-call-of-duty-modern-warfare-2s-source-code-83-percent-of-the-2009-shooters-functions-now-match-the-original-byte-for-byte#main
- https://www.tomshardware.com/membership
- Why Telecom Operators Are Building Their AI Strategy on Open Models
- AMD debuts Helion reference design for agentic AI servers with up to 32,768 Zen 6 ‘Venice’ cores
- Into the Omniverse: How Developers Turn Ideas Into Simulations With Frontier AI Agents
- From Scan to Treatment Plan, AI Helps Close Breast Cancer’s Deadliest Gaps
- From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI
Informational only. No financial advice. Do your own research.