A company's AI read my CV and concluded that my primary professional role was "American football coach and scout."
I am not a professional coach. I have coached American football as a volunteer for years, outside of work. It is on my CV, listed under extracurricular activities. My actual most recent professional role is Head of Digital Factory.
The AI did not make that distinction. It read the CV as a flat list of roles, extracted the one that matched its title-detection logic, and filed it as my job title. Everything downstream was built on that.
What happened next
The company used an AI scoring tool to evaluate my fit for the role. The tool compared my profile against the job's required skills. It found 2 matches out of 21.
That figure is not about me. It is about a fictional candidate whose primary job is coaching a sport. That candidate has no background in digital transformation, platform engineering, or organisational leadership. Naturally, 2 out of 21 skills matched.
I only found out because I submitted a GDPR Article 15 request after my rejection. The company's response included the full output of the AI tool, the extracted job title, the skills comparison, and the score. Without that request, I would have received a template rejection email and nothing else.
Why parsers get this wrong
Resume parsers are trained to extract structured information from unstructured text. They look for patterns that resemble job titles: a noun phrase, a date range, an employer name nearby. They do not understand hierarchy. They do not distinguish between a role that occupied thirty years of a career and one that occupies two evenings a week.
A CV that lists both a Head of Digital Factory role and a volunteer coaching role gives the parser two candidates for "job title." Which one it picks depends on the model's training data, the formatting of the document, and the proximity of surrounding signals. It is not reading intent. It is matching patterns.
The more varied a candidate's background, the higher the risk. A career that includes board memberships, advisory roles, academic positions, or community involvement alongside professional employment gives the parser more options and more opportunity to pick the wrong one.
This is not an isolated case
I have seen the same pattern in another SAR response, where a company's system filed my role as vice-president of a cultural association, an activity I listed as a community involvement. Same mechanism: a title-shaped string extracted from the wrong part of the document, treated as the primary professional identity.
In both cases the candidate data held by the company did not reflect the candidate. In both cases the downstream evaluation was run against that incorrect record. In both cases the candidate had no way of knowing without asking.
What GDPR gives you
Article 15 gives you the right to see what a company holds about you. Article 5(1)(d) requires that data to be accurate. Article 16 gives you the right to request rectification if it is not.
None of those rights help if you do not know the error exists. The SAR is what surfaces it.
What this means for your applications
If a company uses an AI scoring tool and your CV includes activities outside your professional career, the score you were given may not reflect your professional profile at all. You have no way to know from the rejection email. You may have one way to find out.
The underlying problem is not that side activities appear in a profile or a resume. It is that when a third-party parser reads your document, it discards the context you gave those activities and re-infers the hierarchy from raw text. What you labeled as extracurricular becomes a job title candidate, indistinguishable from your professional roles.
The fix is not to remove those activities. It is to know that any system re-parsing your document from scratch may get this wrong, and that the only way to find out is to ask. Job-Agent includes a GDPR Article 15 request template to help you do exactly that.
