AI job-search tools: what they can and cannot do

CV and cover letter writers

A chat assistant will turn notes into a tidy CV in minutes, and that is real help. What it cannot do is know which of your results matter to the person you are writing to. It works from what you give it, so a draft tends to be fluent and general, and the specific claim that would get you a meeting is the part you still have to write.

The bigger limit is where your words go. OpenAI's help page says that on the consumer versions of ChatGPT, your conversations may be used to improve its models by default. You can turn that off in Settings, then Data controls, then Improve the model for everyone. If you rate a reply with a thumbs up or down, the whole conversation may be used for training even with the setting off. The business versions do not train on your content by default.OpenAI Help

Not training on something is not the same as not keeping it. A Temporary Chat is not used for training, but OpenAI says it may keep a copy for up to 30 days for safety purposes.OpenAI, Temporary Chat

• Turn the training setting off, or use a Temporary Chat, before you paste a CV in.
• Take out client names, deal values and anything from your employer's internal documents first. A draft does not need them to be good, and anything you leave out cannot be kept.
• Do not rate the replies while the conversation contains your CV.

Auto-apply tools

These tools submit applications for you, in volume, to any advertised role that matches the filters you set. The appeal is obvious. The limits follow from how they work.

• You stop choosing where your CV goes. A filter does not know where you work now, which companies are clients of your employer, or which recruiter talks to your manager.
• Every application is a copy you have to ask for back. Each one leaves your CV with another system, under that system's own rules for how long it keeps it.
• Applications go out in your name that you have not read. A cover letter written to fit a filter is still signed by you.
• They only reach advertised roles. A tool that reads job boards cannot apply for a role that was never posted.

Some of these tools work by logging in to sites as you. Before you connect one, read the terms of each site it will act on: a tool using your account is, as far as that site is concerned, you.

Keyword scorers and CV optimisers

These compare your CV with the words in a job advert and give you a match score. A score tells you how much your wording overlaps with one advert. It does not tell you how a particular employer's system is set up, because no tool outside that employer can see it, and it says nothing about whether the person reading the shortlist will want to meet you. Use it to catch a missing term, not to rewrite your record around an advert.

The tools on the other side

Employers and recruiters use AI too, and what it does with your information is usually invisible to you. In November 2024 the UK's Information Commissioner's Office published what it found when it audited providers of AI sourcing, screening and selection tools.ICO, Nov 2024

• Several providers inferred people's gender and ethnicity, often from their name, rather than asking them.
• Some tools collected far more personal information than they needed. In some cases it was scraped from millions of profiles on job networking sites and social media to build databases recruiters could search.
• Some search features allowed recruiters to filter out candidates with certain protected characteristics.
• Several providers with large candidate databases intended to keep the information indefinitely.ICO, Nov 2024

The ICO made 296 recommendations across those audits. In the UK you also have rights when a decision about you is made by automated means alone and has a significant effect, such as an online test that sifts applications with no person involved. The ICO says the organisation should let you ask for a person to step in, give your view and challenge the decision, and that you can ask for the reasons. It notes that this guidance is under review after the Data (Use and Access) Act 2025.ICO, Nov 2024ICO, automated decisions

Separately, you can ask any organisation whether it holds personal information about you and for a copy of it, and it usually has one month to respond.ICO, access requests

In New York City, an employer using an automated employment decision tool must have had it bias-audited within the past year, publish a summary of the results, and tell candidates who live in the city at least 10 business days before using it, including what qualifications and characteristics it will assess. The EU's AI Act classes AI systems used to place targeted job adverts, filter applications or evaluate candidates as high-risk.NYC DCWPEU AI Act

One test for every tool

1. Where does my CV go when I use this, and who decides?
2. Who keeps a copy, for how long, and can I have it deleted?
3. Does anything go out in my name that I have not read?
4. If my employer saw what this tool did, would I be comfortable?

A tool that answers all four well is worth the time it saves. One that cannot answer the first is the fastest way to turn a quiet search into a public one.

Overture's own answers

Overture is an AI career tool too, so the same four questions apply to it, including where the answer is not a clean no. These answers are from its privacy policy, its terms and its pricing page.

1. Where does my CV go when I use this, and who decides? Into your account, where Overture uses it to research where you could go next. It never applies for you, and nothing reaches an employer unless you approve an approach. Some of the work is done by providers it names: hosting, transcription of your interview audio, identity checks, email delivery, payments and product analytics, each bound to use your data only for that job.
2. Who keeps a copy, for how long, and can I have it deleted? Overture, while your account is open. Close it and your personal data is deleted within 30 days, except a record of a transaction that must legally be kept. Your material is never sold and never used to train models that serve anyone else.
3. Does anything go out in my name that I have not read? Yes, in one case. You read and approve the first note of an approach. Approving it also allows up to three short follow-ups on days 3, 7 and 14, and those are written when they go, so you may not see them first. A reply stops them, and you can stop all sending at once, at any time.
4. If my employer saw what this tool did, would I be comfortable? Companies on your excluded list are never researched or approached, and the list is checked before anything runs. An approach goes in your own name, from an Overture address set up for you, and its first note asks the reader to keep your search discreet. Whether they do is up to them.Overture privacyOverture termsOverture pricing

Questions people ask

Can I stop ChatGPT using my CV to train its models?

Yes. OpenAI says you can switch off Improve the model for everyone under Settings, then Data controls, or use a Temporary Chat. Its business products do not train on your content by default. Rating a reply can still put that conversation into training, and a Temporary Chat may still be kept for up to 30 days.

Does an employer have to tell me it uses AI to screen applications?

It depends where you are. In New York City, yes: candidates who live there must be told at least 10 business days before an automated tool is used. In the UK, if a decision about you is made by automated means alone, the ICO says you can ask for a person to look at it, give your view and challenge it, and you can always ask an organisation what personal information it holds about you.