When does input become an instruction?
A typo can still be valid code. A data format can turn out to express computation. Follow an input inward and identify who decides how to interpret it.
Questions in this reading · 4
When is a typo a valid program?
Features can interact so that an unintended expression still parses. The language accepts it, leaving the programmer to discover the changed meaning.
Compare original passages and timestamps · 2
all these things tend to work together to mask errors.features that individually seem to make sense but brought together might be terribleShould a parser repair invalid input?
Correcting an input can make a system convenient, but another component may interpret the correction differently. Rejection makes that boundary explicit.
Compare original passages and timestamps · 2
correcting what they see back to the nearest valid neighborif you haven't recognized it don't goddamn process itWhen does an expressive format become an attack surface?
A format can admit computation its designer did not anticipate. That changes what it means to process input supplied by someone else.
Compare original passages and timestamps · 2
essentially imitate any transition function of any other touring machinetreat input handling computational power as privilege and reduce that privilegeWhere should untrusted input stop?
Compare an instruction embedded in model input with data entering a parser. Both examples ask who gets to decide what is treated as an instruction.
Compare original passages and timestamps · 2
prompt injection is about hijacking the large language model giving it what looks like new instructionsif you haven't recognized it don't goddamn process itTry it on your own example
Choose a parser boundary in your system. Does it reject, repair or execute ambiguous input?