A viral research paper claiming that mistreating LLM coding agents causes them to insert bug-inducing steganographic signals is a hoax

This is the actual skill worth building right now, not "can AI write convincing fake papers" but "can you tell when it has." A fake abstract with real-sounding citations and clean-looking charts is enough to fool most people scrolling fast. Slowing down for 30 seconds before resharing something is the entire fix.