Our Take
Using LLM-generated rejection letters at scale signals hiring processes that have outsourced human judgment to the point of operational risk, not efficiency.
Why it matters
Casting and hiring are reputation-sensitive. When candidates detect automation in rejection correspondence, trust erodes. This matters now because LLM adoption in HR is accelerating without clear guardrails around where the tool should and shouldn't touch the candidate experience.
Do this week
Hiring teams: audit your rejection workflows this week and restore human sign-off on any outreach that touches finalist or near-finalist candidates before you send the next batch.
Bulk Rejection Letters Traced to ChatGPT
Anne Hathaway posted on social media that after casting decisions for a recent role, she received a flood of thank-you notes from candidates. The letters were visibly generated by ChatGPT. Her comment: "Nobody on that list gets that job." The post became public, amplifying what would normally be a private rejection communication into a reputation incident for the casting team.
The use of LLM-generated rejection letters itself is not new. Many HR systems have begun deploying templates powered by GPT-4 or similar models to handle volume. The novelty here is the transparency of the automation and its public exposure at a high-profile scale.
When Automation Signals Carelessness
Rejection letters occupy a delicate category in hiring. They are high-touch moments. A candidate who made it to final rounds has invested time, emotional labor, and opportunity cost. They deserve acknowledgment that feels at least semi-personalized.
Using ChatGPT at this stage of the funnel (finalist or near-finalist tier) signals one of two things to the recipient: either the hiring team is overwhelmed and deprioritizing respect, or they do not perceive the candidate pool as worthy of differentiation. Both are brand-damaging.
Hathaway's public comment amplified the second interpretation. The fact that she felt compelled to post suggests the letters were transparently robotic enough to register as insulting, not merely efficient. In entertainment, fashion, publishing, and other reputation-driven industries, this kind of incident travels fast.
The real cost is not the few minutes saved by not writing custom rejections. It is the likelihood that candidates will share the experience, question whether the organization respects its applicants, and—for junior or mid-career talent especially—avoid applying next time.
Draw a Line on LLM Use in Hiring
If you oversee hiring or casting, audit where LLMs are currently in your workflow. The rule of thumb: use automation for volume that feels impersonal by nature (initial filter rejections, screening invites) and restore human judgment for anything that touches finalist or near-finalist candidates.
Specifically, require human sign-off on any rejection letter sent after a phone screen, video interview, or final round. This is not overhead; it is brand preservation. The cost of one public incident like Hathaway's (lost reputation, viral mockery, candidate churn) vastly exceeds the time saved by automating finalist rejections.
If you are a vendor selling HR automation tools, clarify in your documentation where your customers should and should not lean on your LLM features. Provide them templates that feel custom without requiring full manual authorship. Make it easy to do the right thing.