The Next Wave of AI Slop is 'Reverse Engineering'

Summary

The speakers describe the growing use of AI tools to clone, decompile, or recreate existing software—and to modify hardware by installing replacement software. They compare this trend to the spread of AI-generated content: some projects may be useful, but many could be unreliable, unsafe, or derivative.

Main Examples and Technology

• Software clones: An individual reportedly used AI to create semi-working alternatives to Adobe Creative Cloud applications and Microsoft Office. A speaker who tried the Adobe-style apps said they opened and looked familiar, and that basic functions, such as editing a photo, appeared to work. The testing was limited, however, and the apps were deleted because of concerns about downloading suspicious software.
• Modifying devices: People are using AI to help replace or alter firmware on smart TVs, removing features such as advertising, tracking, and unwanted software. Other reported targets include Lululemon workout mirrors, a Meta tablet, inexpensive smart photo frames, Peloton bikes, baby-monitor software, and Ableton Move.
• Reviving unsupported hardware: AI-assisted work may help restore devices abandoned by their manufacturers, bypass subscription restrictions, or unlock features that the hardware can already support. The speakers connect this to the broader “right to repair” movement and cite John Deere tractors, where software restrictions can distinguish differently priced versions of similar hardware.
• Games and other software: AI tools are being directed at video games and individual features—for example, examining the Chrome offline dinosaur game to understand and alter its behavior. The discussion also mentions claims of reverse-engineering guitar effects pedals, though the speaker did not independently verify that example.

The GitHub Project

The speakers discuss a popular GitHub repository called “Reverse Engineering Anything.” It is described not as an AI model itself, but as a collection of tools and instructions that helps existing AI agents inspect software such as native binaries, JavaScript and Electron apps, .NET assemblies, and websites. The analysis is performed locally, and the tool is said to provide evidence and limitations for its conclusions.

The repository may lower the skill and effort needed to investigate software or hardware. Tasks that once required specialist knowledge and substantial manual work can now be attempted by asking an AI agent to guide the process, though users may still need to complete some steps themselves.

Risks and Broader Implications

• Results vary substantially: some may closely reproduce the original, while others may be buggy, incomplete, or assembled from unrelated components.
• Installing untrusted software or firmware could expose devices to malware, data theft, security flaws, or operational failures.
• AI-generated modifications could make their way into products sold to others, potentially creating “garbage devices” with poorly tested or stolen software. The speakers speculate that this could affect businesses and public infrastructure as well as personal gadgets.
• Software developers may face competition from AI-produced clones, even when those clones are lower quality.
• The discussion raises unresolved intellectual-property questions. It compares copying the behavior or sound of a guitar effects pedal to disputes over guitar designs, noting that the legal treatment of a product’s “sonic structure” could be difficult to determine.

Reviews, Guides, and Tutorials

• Review: A speaker briefly tested Adobe-style AI clones. They appeared recognizable and supported basic use, but the review was not thorough; the speaker deleted the apps because of security concerns.
• Guide/tutorial: No step-by-step tutorial is provided. The GitHub repository is described as a toolkit that can guide AI agents through reverse-engineering tasks, with users potentially needing to complete some setup and device-specific steps.

Main Speakers and Sources

• Jason and Sam, speaking on the 404 Media podcast.
• The discussion also refers to reports and online claims from GitHub, X, Hacker News, and device-modification communities; some individual claims are presented as unverified.