The term entered widespread use after the AntiPatterns book was published. The authors were referring to "patterns of failure" commonly seen in software projects.

Nope, the AntiPatterns book defined it as something like "recurring patterns of failure, which look superficially attractive".

Edit: yes, I am extremely well aware of the conventional definition. I was deliberately offering an alternative spin in context to the parent.

Case in point: ORM can handle 80% of query plans (patterns). Some, it can't handle. They are not patterns, they are the opposite: unique and non replicable. That doesn't make them bad.

However, I disagree with headline. Just because ORM doesn't handle non pattern situations doesn't make the concept an anti pattern.