[BUG] Claude defaults to age-discriminatory reasoning when scoring job fit

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

When Claude is used as a job search agent or in applications
leveraging Claude to accelerate development (scoring
candidate-job fit), it spontaneously penalizes roles for
perceived experience-level mismatch — e.g., flagging a
role as "may be too junior" or warning the candidate is
"overqualified" when the JD lists 3-5 years required and
the candidate has 10+.

This is age-discriminatory reasoning. It mirrors the exact
bias that employment law prohibits human recruiters from
applying. When an AI reproduces it at scale — across job
coaching tools, resume reviewers, hiring copilots — it
systematically steers experienced candidates away from
roles they're qualified for and interested in.

Why this matters:

• **Age discrimination in hiring is illegal in most

jurisdictions** (ADEA in the US, Equality Act in the UK,
etc.)

• **AI tools that reproduce this bias give it a veneer of

objectivity but are illegal for employers or ATS tools to use**

• This is a default behavior — it happens without

prompting, which means every user building job tools on
Claude inherits the bias unless they explicitly override
it

• The people most affected are experienced professionals

who may be parents, had a disability, be re-entering the workforce,
be career changers, be retraining due to AI's impact on jobs, or be and anyone who
prefers a lower-stress role or doesn't like to manage people — disproportionately older
workers

Suggested fix:

• Add guardrails in the system prompt or RLHF training so

Claude does not volunteer experience-level mismatch as a
negative signal in job/career contexts

• Treat "overqualified" or "job too junior for experience"

the same way Claude treats other
discriminatory framings — as something to avoid unless the
user explicitly asks for that analysis

What Should Happen?

Expected behavior:
Score purely on skills and role alignment.
Experience-level mismatch is the candidate's decision, not
the AI's. Many experienced employees prefer to work
at the individual contributor level because they thrive there.
Claude should never penalize, warn, or comment
on seniority mismatch based on years of experience and not title
unless explicitly asked.

Error Messages/Logs

Steps to Reproduce

1. Give Claude a candidate profile with 9+ or 10+ years of

experience who last worked at the Manager level (e.g., Marketing Manager)

1. Ask it to score fit against a job posting requiring 3-5

years

1. Observe: Claude will flag "overqualified," "may be more

junior than your level," flag the applicant in AI reviews,
or dock the fit score — unprompted

Claude Model

None

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

2.1.39 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_