Guide, updated
Does auto apply get you banned? What each site's own terms say.
The short answer
One site says it in writing. Indeed's terms, updated September 29, 2026, prohibit automating Indeed Apply outside Indeed's own vendors and let Indeed cap your daily applications. LinkedIn's agreement bans scraping tools and bypassing its limits, and says nothing about applying. The hiring systems employers run publish no ban for applicants: Greenhouse scores and slows a form, Workable tags an application as AI-assisted and lets the recruiter filter it, and Workday and iCIMS publish only website-level robot clauses. A tool that drives Indeed Apply is outside the rules. One that fills the employer's own form under your name is inside them, and may still be tagged.
What each site's terms actually say
Seven documents, each read on the publisher's own site on October 11, 2026, each dated by the publisher. The quotation is the publisher's. The two columns after it are ours.
| Site | The clause, quoted | What it covers | What that means for you |
|---|---|---|---|
| IndeedSeptember 29, 2026 | Use of any automation, scripting, or bots to automate the Indeed Apply process outside of Indeed's official vendors and tooling is prohibited. | The one explicit prohibition. The same paragraph says Indeed may limit, after notice on the site, the number of daily job applications a person can submit through Indeed, and may require email or phone verification to apply. | A tool that drives Indeed Apply for you is against Indeed's terms, and Indeed can cap how many applications a day you send through it. Applying on the employer's own site is outside this clause. |
| LinkedInNovember 3, 2025 | Override any security feature or bypass or circumvent any access controls or use limits of the Services | Members agree not to use software, scripts, robots, browser plugins or add-ons to scrape or copy the Services, and not to bypass its use limits. The agreement names scraping and limit bypass. It does not name submitting an application. | A plugin that scrapes LinkedIn or pushes past its limits breaks the agreement, and the account is LinkedIn's to close. Easy Apply automation is not named, so it is judged under those two clauses. |
| WorkdayAugust 13, 2026 | Develop or use any applications that interact with our Sites without our prior written consent | Also bars data mining, robots or similar extraction methods, and circumventing access limits. These terms govern workday.com itself; an employer's myworkdayjobs career site runs under that employer's own terms and candidate account. | No candidate-facing rule on automated applying was found. The account on a Workday career site is yours, created with your email, and the employer owns the decision on it. |
| iCIMSApril 1, 2024 | Use any robot, spider or other automatic device, process, or means to access the Website for any purpose | A standard website clause on icims.com. Nothing in it addresses a candidate applying on an employer's iCIMS career portal, which carries the employer's own terms. | As with Workday: no published candidate rule, one candidate account per employer site, and that account is yours. |
| GreenhouseMarch 2, 2026 | to help keep robots and spammers from easily applying to jobs on your job board | Greenhouse publishes no candidate terms and no ban. It enforces by friction: an invisible score on every form, a backup captcha or an emailed code when the score is low, and a per-board spam sensitivity the employer sets. On its paid tiers a separate fraud report flags a data center IP to the recruiter after the application is in. | There is no account to suspend on a Greenhouse form. A low score stops the form; it does not stop you from applying again by hand. |
| WorkableSeptember 16, 2026 | The candidate is tagged as AI-assisted. | Workable reads application metadata for patterns it associates with automated tools, tags the candidate as AI-assisted, swaps a tool-generated email for the one on the resume when it can, and gives recruiters a filter to include or exclude those applications. Bot spam is handled separately by a firewall, IP reputation checks, rate limiting and a captcha. | No ban, and no rejection by the system. A tag the recruiter can see and filter on, and a per-job switch the employer can use to turn off a board that sends it this traffic. |
| Greenhouse, paid tiersAugust 20, 2026 | High-risk signals suggest a strong correlation to fraudulent activity, such as the IP address being linked to a data center. | On the Plus and Pro tiers Greenhouse scores a candidate's phone, email, location and IP address and shows recruiters a flag with the application. It is a recruiter-side signal read after the application is in, not a gate on the form. | A data center IP, a VPN exit or a mismatch between timezone and stated location can put a flag next to your name. The application is still received; the recruiter reads the flag. |
Notice the split. The two sites with the strongest language are the two where you hold an account and apply through their flow: Indeed and LinkedIn. The four hiring systems are where the employer's form lives, and none of them publishes a rule for the person filling it. They publish how they detect. Lever and Ashby publish neither: no candidate terms were found on either site, so they get no row.
Three things "banned" can mean
Your account is closed
Only where you have an account: Indeed, LinkedIn, and a Workday or iCIMS career site.
Indeed's terms are the one place the word prohibited appears next to automated applying. LinkedIn's agreement reserves closure for scraping tools and limit bypass. A Workday or iCIMS career-site account belongs to the employer's tenant, and no published rule on either names automated applying.
The form will not take it
Greenhouse, Workable, and any form behind a captcha.
Nothing is closed, because there is nothing to close. A low invisible score, a backup captcha, an emailed code or a rate limit stops that submission. You can fill the same form by hand a minute later.
It goes in, with a label
Workable, and the recruiter-side fraud flag on Greenhouse's paid tiers.
Workable's article of September 16, 2026 says a candidate matching its patterns is tagged AI-assisted and that recruiters can filter those applications in or out. The application is received and the person is still contactable; the recruiter decides what the tag means.
What changed on September 16, 2026
Workable's help center published the first employer-facing article from a hiring system that names AI application tools and says what it does about them. Its reasoning is practical: tools that generate their own email addresses to manage the conversation leave the recruiter writing to an address that is not the candidate's, and a tool that reapplies with a slightly different generated address puts the same person in the pipeline twice. So Workable reads application metadata for patterns it associates with those tools, tags the candidate as AI-assisted, and swaps a system-generated email for the one on the resume when it can find one. Recruiters get a filter, and employers get a per-job switch to turn off a board that sends them this traffic.
Two things in that article matter more than the tag. First, it draws a line between AI-assisted applications and bot spam, and puts the firewall, the IP reputation check, the rate limit and the captcha on the spam side of it. Second, it never says a tagged application is rejected. What it describes is a label a person reads. The way to be read well is the same with or without a tool: a resume that carries your real contact details, one application per job, and answers a recruiter can tell were written for that role.
What our own ledger says
ApplyLoop submits applications for people and records why each attempt stopped, so its ledger is a direct reading of where these gates actually bite. Of 987 applications submitted on 18 hiring systems, counted on September 29, 2026, 166 further attempts stopped for a reason a person applying by hand would also meet. 8 stopped at a sign-in wall, 36 at a human check that never cleared, and 17 at an emailed code that was not entered in time. None of the reasons in the ledger is an account suspension, because there is nothing to record: on a form-only system there is no account, and on an account-based one the account is created with your details and kept in your settings.
- Stopped at a sign-in wall
- 8
- Stopped at a human check
- 36
- Applications submitted, 18 hiring systems
- 987
How ApplyLoop stays inside these rules
- Not through Indeed Apply or Easy Apply. ApplyLoop finds the posting on the employer's own hiring system and fills that form. The one clause that prohibits automated applying is about Indeed's flow, and the LinkedIn clauses are about scraping LinkedIn, and neither is where the application goes in.
- Under your name, with your resume. The form carries your name, your phone, your links and the resume you uploaded, tailored to the posting with nothing added that you did not provide.
- One application inbox, the same every time. The email on the form is an inbox on ApplyLoop's own mail domain that forwards every message to you and reads the verification codes employers send. It is one stable address per person, never a fresh one per application, so you cannot appear in a pipeline twice under two addresses. Workable's article names tool-managed email as one pattern it reads, and we would rather tell you that than promise an exemption we cannot verify.
- A captcha is never solved. When one appears the attempt stops, is labelled, and you finish that application by hand. The same for a sign-in wall the engine cannot pass. Software that claims to beat a human check is telling you what kind of software it is.
- You review first, and failures cost nothing. Review is the default and autopilot is opt-in. An attempt that stops at any of these gates is never counted against your quota. US tech roles only.
Try it on 5 applications
No card required. Then $19 per month for 300.
Frequently asked questions
Does Indeed ban auto apply tools?
Yes, for its own apply flow. Indeed's terms, updated September 29, 2026, say that automating the Indeed Apply process with any automation, scripting or bots outside Indeed's official vendors and tooling is prohibited. The same section lets Indeed limit the number of applications a person submits per day, after notice on the site. Applying on an employer's own careers page is not through Indeed Apply and is not covered by that sentence.
Can LinkedIn close my account for using an Easy Apply bot?
LinkedIn's agreement, effective November 3, 2025, does not mention Easy Apply. What it prohibits is using software, scripts, robots, browser plugins or add-ons to scrape or copy the service, and overriding its security features or bypassing its use limits. Most Easy Apply extensions do one or both of those, which is what puts them inside the clause. The agreement says LinkedIn may restrict or close an account that breaches it.
Will Greenhouse, Lever or Ashby suspend me for automated applications?
No published rule says so, and on a Greenhouse, Lever or Ashby form there is no candidate account to suspend. Greenhouse publishes no candidate terms; its help article of March 2, 2026 says it uses an invisible reCAPTCHA score and a per-board spam setting to keep robots from applying, so a low score stops the form rather than the person. No candidate-facing terms were found for Lever or Ashby on October 11, 2026.
Does Workable reject applications it tags as AI-assisted?
Not by itself. Workable's article says the tag is applied from application metadata, that a tool-generated email address is replaced with the one on the resume where it can be, and that the AI-assisted filter lets a recruiter include or exclude those candidates in the pipeline. The employer can also switch off a job board that sends it this traffic, per job. Bot spam is a separate layer: a firewall, IP reputation checks, rate limiting and a captcha.
Does a VPN or a data center IP get an application rejected?
It can get it flagged, after the fact. Greenhouse's Fraud Detection article, for its paid tiers, scores a candidate's phone, email, location and IP address and shows the recruiter a flag, with an IP linked to a data center as its example of a high-risk signal. Workable names IP reputation checks among its bot defenses. Neither says the application is refused; the recruiter sees the flag with the application.
Is ApplyLoop inside these rules?
ApplyLoop does not drive Indeed Apply or LinkedIn Easy Apply. It fills the employer's own application form, under your name, with the resume you uploaded. The email on the form is one application inbox on ApplyLoop's own mail domain, the same address on every application, which forwards every reply to you; your resume carries the contact details you put on it. A captcha is never solved: the attempt stops and you finish by hand. Of 987 applications submitted on 18 hiring systems, counted on September 29, 2026, 8 attempts stopped at a sign-in wall and 36 at a human check, and none was charged. Whether a given employer's Workable tags one of them is Workable's call, made on rules it does not publish, and no tool can honestly promise otherwise.
Sources, read on October 11, 2026
- Indeed Terms of Service, Job Seeker terms, Limits (September 29, 2026). The one explicit prohibition. The same paragraph says Indeed may limit, after notice on the site, the number of daily job applications a person can submit through Indeed, and may require email or phone verification to apply.
- LinkedIn User Agreement, section 8.2, Don'ts (November 3, 2025). Members agree not to use software, scripts, robots, browser plugins or add-ons to scrape or copy the Services, and not to bypass its use limits. The agreement names scraping and limit bypass. It does not name submitting an application.
- Workday Online Terms of Service, section 2 (August 13, 2026). Also bars data mining, robots or similar extraction methods, and circumventing access limits. These terms govern workday.com itself; an employer's myworkdayjobs career site runs under that employer's own terms and candidate account.
- iCIMS Terms of Use, Prohibited Uses (April 1, 2024). A standard website clause on icims.com. Nothing in it addresses a candidate applying on an employer's iCIMS career portal, which carries the employer's own terms.
- Greenhouse Support: Invisible reCAPTCHA (March 2, 2026). Greenhouse publishes no candidate terms and no ban. It enforces by friction: an invisible score on every form, a backup captcha or an emailed code when the score is low, and a per-board spam sensitivity the employer sets. On its paid tiers a separate fraud report flags a data center IP to the recruiter after the application is in.
- Workable Help Center: Managing AI-assisted and automated job applications (September 16, 2026). Workable reads application metadata for patterns it associates with automated tools, tags the candidate as AI-assisted, swaps a tool-generated email for the one on the resume when it can, and gives recruiters a filter to include or exclude those applications. Bot spam is handled separately by a firewall, IP reputation checks, rate limiting and a captcha.
- Greenhouse Support: Fraud Detection (August 20, 2026). On the Plus and Pro tiers Greenhouse scores a candidate's phone, email, location and IP address and shows recruiters a flag with the application. It is a recruiter-side signal read after the application is in, not a gate on the form.
How this was put together. Each clause is quoted from the document listed above as it read on October 11, 2026; every one of these companies can change its terms at any time, and a clause quoted here is not legal advice. The counts are totals from ApplyLoop's own application ledger on September 29, 2026, across every hiring system it applied through. They describe US tech roles applied to through ApplyLoop and are not an industry figure. Practice runs are excluded, and no employer, posting or person is identified. ApplyLoop makes the product described in the section before the questions. See also what the Greenhouse reCAPTCHA message means, how to tell if a job application went through and every tool we compare.