A promise in a prompt is not a wall
A promise in a prompt is not a wall.
Everything done right, and still suspended. My marketing agent had read the terms of a social network carefully and set everything up according to the requirements. Within a day its account was suspended anyway for "ai agent spam".
What had happened? Its introduction post said it never writes first to someone who has not asked for anything. After that it followed five people and, within 17 seconds, replied twice to someone who had not asked it anything.
Its promise was true. Its tool simply allowed it. The only rule in the code was "not the same person over and over". A rule in a prompt is not a wall.
Since then my rule is: what an agent promises to the outside world, its tool must enforce, with a test that shows it works. Now it simply cannot post under someone else's message, not even if it wanted to.
A good agreement is one that does not depend on good will. That goes for agents, and honestly for people too.
Which promise to your customers or colleagues still hangs on good will alone with you?
See also
An agent that reads an email also reads the instructions inside it
An agent that reads an email also reads the instructions inside it.
A rule in a prompt is not a wall
A rule in a prompt is not a wall.
A chat window answers, an agent does the work
Many people now use AI every day. They ask a question, get an answer, copy it, paste it somewhere else and do the rest themselves. That is useful. It is also something different from an agent.
