The new DDoS: Unicode confusables can't fool LLMs, but they can 5x your API bill Can pixel-identical Unicode homoglyphs fool LLM contract review? I tested 8 attack types against GPT-5.2, Claude Sonnet 4.6, and others with 130+ API calls. The models read through every substitution. But confusable characters fragment into multi-byte BPE tokens, turning a failed comprehension attack into a 5x billing attack. Call it Denial of Spend.
Every V86 segment gets the same treatment: access rights 0xE200 (Present, DPL=3, writable data segment), base = selector shifted left by 4, and limit = 64 KB. The microcode loops through all six segment register caches using a counter, applying the same fixed descriptor to each one. This is pure real-mode emulation, enforced at ring 3 with full paging protection underneath.
。夫子对此有专业解读
What is chickenpox and who can get an NHS vaccine?
Get our breaking news email, free app or daily news podcast