Indemnity Clause in a Contract: Reading the Liability Cap
Here is a sentence that turns up in client-drafted contracts often enough to be a template: "Contractor shall indemnify, defend, and hold harmless Client from any and all claims, damages, losses, and expenses arising out of or related to Contractor's services." No cap. No carve-out. No mention of what happens if the claim traces back to something the client did, not the contractor. On a $6,500 branding project, that one sentence attaches an unbounded amount of someone else's legal exposure to the freelancer who signs it, for as long as anyone can argue the word "related" reaches far enough.
Most freelancers skim past this pair of clauses because neither one describes work. There's no deliverable, no deadline, no dollar figure attached to a milestone — just two paragraphs, usually near the back, that decide what happens if something goes wrong for someone who isn't even a party to the contract. That's exactly why they're worth reading slowly. Indemnification and the liability cap are the sentences that decide how much money can travel back out of your pocket, as opposed to every other clause on this site, which is about getting money to travel in.
None of what follows is legal advice, and it can't tell you what a court in your state would do with either clause. What it can do is show you the three moving parts inside "indemnify," the four shapes a liability cap usually takes, and which exclusions from that cap are ordinary versus which ones quietly undo it.
What "indemnify and hold harmless" actually promises
Strip the Latin and the clause is doing three separate jobs, and contracts don't always spell out all three.
Who pays. Indemnification is triggered by a third party — someone outside the contract. A client alleging you delivered late is not a third party; that's a breach-of-contract dispute between the two of you, and it's the liability cap that governs it, not indemnity. A photographer whose stock image you used without a license suing the client is a third party, and that's what indemnification is for. Mixing the two up is the single most common misreading of this section, and it matters because the liability cap almost never applies to indemnity obligations — more on that below.
When the money moves. Well-drafted indemnity clauses distinguish a duty to defend from a duty to indemnify, and the two run on different clocks. The duty to defend can arise the moment a claim is merely filed and alleges something the clause covers — you're paying for a lawyer before anyone has decided who's actually at fault. The duty to indemnify arises later, once fault and damages are established by judgment or settlement. The American Bar Association's construction-law bulletin on this split — where the distinction gets tested constantly, because indemnity fights over defense costs are common in that industry — describes the defend duty as broader and earlier-triggered precisely because it doesn't wait on the outcome (americanbar.org, Under Construction archive, checked 25 September 2026). If your contract's sentence reads "defend, indemnify, and hold harmless," assume both duties are live and ask who controls the defense once it starts.
Who's in charge once a claim lands. This is the part almost nobody reads on a first pass. Look for language saying the indemnifying party has "sole control of the defense and all related settlement negotiations." That's standard, and it's supposed to be standard — whoever is paying should get to choose the lawyer and the strategy. What you want to confirm is that a settlement can't bind you without your written consent if you're the one footing the bill, and that the other party has to give you prompt written notice of the claim. A notice requirement sounds like paperwork; in practice it's the difference between finding out about a lawsuit against you in week one, when you can still get ahead of it, or in week twelve, after the client's lawyer has already made representations on your behalf.
One-way indemnity is the sentence that should stop you
Read the clause once for direction. Does the contractor indemnify the client, does the client indemnify the contractor, or both?
A one-way clause running only from you to the client is the version that shows up in client-drafted paper by default, and it's worth pushing back on regardless of how small the job is. The AIGA Standard Form of Agreement for Design Services (2022 update, checked 25 September 2026) drafts this section as two mirrored obligations. Section 10.1 has the client indemnify the designer for claims arising from the client's own breach of its responsibilities, warranties, or obligations. Section 10.2 has the designer indemnify the client for claims that are "inconsistent with Designer's representations and warranties" — and it expressly carves out anything caused by client content, third-party materials, unauthorized modifications, or the client's own gross negligence or misconduct. That carve-out is the part client paper drops most often. Without it, you can end up indemnifying a client against a claim that only exists because they changed your file after delivery, or ignored your written warning about a licensing restriction.
The fix is not to delete the clause — clients reasonably want protection when a freelancer actually breaches a warranty, misses a license, or is negligent. The fix is to make it mutual, and to tie your side of it to your own breach, your own content, and your own negligence, rather than to "any and all claims . . . related to the services," which is broad enough to cover a claim that's really the client's fault.
One more thing worth noticing while you're in this section: the same AIGA document's notes on warranties recommend limiting a designer's promise of non-infringement to "the best of Designer's knowledge," rather than an absolute guarantee, because trademark and patent rights — unlike copyright — can be infringed by accident, with no copying involved at all (AIGA Standard Form introduction, pp. 13–14, checked 25 September 2026). If your warranty section makes an unqualified promise that your work doesn't infringe anyone's rights, and your indemnity section is also uncapped, you've stacked an absolute promise on top of unlimited exposure for a single sentence you signed without reading twice. Loosening the warranty to "best of knowledge" is usually an easier ask than capping the indemnity, and it does some of the same work.
The cap, and the four shapes it comes in
Once you've sorted indemnity, the liability cap is the sentence that limits what either side can recover from the other for breach of the contract itself — the ordinary stuff, not third-party claims. It shows up in roughly four forms, and they are not equivalent.
No cap at all. The contract is silent, which doesn't mean nothing applies — it means the general rule of contract damages applies instead, and in most US states that rule allows recovery of losses that were reasonably foreseeable at signing. Foreseeable damages for a botched product launch, a missed ad-campaign date, or a broken e-commerce checkout can run to a large multiple of what you were paid. Silence favors whoever has the bigger claim, and it's rarely you.
Capped at the total fee. This is the version the AIGA form uses: "the maximum liability of Designer . . . shall be limited to the total Project fee of Designer" (Section 10.4, checked 25 September 2026), paired with an outright exclusion of indirect, incidental, special, consequential, exemplary, and punitive damages "even if Designer has been advised of the possibility of such damages." That last phrase matters — it's there because without it, a client could argue that once they'd warned you a delay would cause specific downstream losses, the cap no longer protects you from those losses. Asking for both halves — the fee cap and the consequential-damages waiver — is asking for a documented, widely used pair, not an unusual concession.
Capped at a multiple, or at fees over a trailing period. For ongoing engagements rather than a single fixed-fee project, a flat "total project fee" cap doesn't map cleanly onto the relationship, because there's no single project fee to point to. Published open-source contract templates aimed at services and SaaS relationships instead cap liability at the fees paid or payable in the twelve months before the claim, sometimes at a stated multiple of that figure (Common Paper's public explainer of this pattern is a useful plain-language reference, checked 25 September 2026). If you're on retainer, this trailing-twelve-months structure is the one to ask for — a total-fees-ever cap on a three-year retainer effectively grows the cap every month you keep working, which nobody drafting the clause actually intended.
A flat number with no relationship to the deal. Occasionally a client's template caps liability at a round number — $50,000, $100,000 — inherited from a much larger engagement and never adjusted down. On a small project this can look generous, but check whether it's also excluding consequential damages; a high-but-uncapped-for-certain-categories number is worse than a low cap that actually excludes the categories that produce runaway numbers.
What should stay outside the cap — and what shouldn't
Almost every liability cap comes with exceptions, and the exceptions are where the real negotiation happens, because a cap riddled with carve-outs isn't really a cap.
Reasonable exclusions from the cap, seen across most standard forms: a breach of confidentiality obligations, gross negligence or willful misconduct, and indemnification obligations themselves (since indemnity is meant to cover potentially large third-party claims, capping it at your fee would often make the promise meaningless to the client). A carve-out for IP infringement claims caused by your own original work — as opposed to client-supplied content — is also common and defensible, provided it only reaches infringement you actually caused.
The carve-out to push back on is one that excludes "any breach of this Agreement" from the cap, which is not a carve-out at all — it's the whole cap deleted through the back door, since every dispute is by definition a breach of the agreement. If you see that phrasing, ask the client what specific risk they're trying to protect against; usually the honest answer is confidentiality or IP, and you can narrow the carve-out to that instead of accepting a clause that reads like a cap but functions like one.
The certificate of insurance funds the promise — it doesn't replace it
Clients who ask for both an indemnity clause and a certificate of insurance are usually trying to make sure there's money behind whichever obligation gets triggered, and that's a reasonable pairing. What it isn't is a substitute for reading the indemnity language, because the two documents answer different questions. The contract says who owes what. The certificate is evidence that an insurer would pay a covered claim on your behalf, up to the policy limit, subject to the policy's exclusions — and a policy limit lower than an uncapped indemnity obligation leaves the gap sitting with you personally. The mechanics of what a certificate actually proves, and the difference between being named a certificate holder and being named an additional insured, are covered in more detail in the piece on W-9s, certificates of insurance, and vendor onboarding forms — it's worth reading both clauses together before you sign, since a client who demands a $2 million COI while leaving the indemnity clause uncapped is telling you, in two separate documents, how large a number they're actually worried about.
A short redline, not a rewrite
You don't need to renegotiate the whole section to fix most of what's wrong with it. Four changes cover the common failure modes: make the indemnity mutual and tie your side to your own breach, negligence, or content rather than to any claim "related to" the services; add a cap set at the total fee for a fixed-price project, or fees paid in the trailing twelve months for a retainer; add the standard exclusion of indirect and consequential damages "even if advised of the possibility"; and check that the only things carved out of that cap are indemnification, confidentiality, gross negligence, and IP infringement you actually caused — not "any breach of this Agreement." None of those four asks requires the client to absorb new risk they weren't already carrying; they mostly ask the client to carry the risk they created themselves, instead of routing it to you through a sentence you might not have read twice.
If the numbers involved are large enough that this section could end your year rather than dent it, that's the point to bring in a lawyer licensed where the contract says disputes are heard — the sentence is short, but what it decides isn't. The nine-clause map of a freelance contract, including where this section usually sits relative to termination and IP, is in the piece on the nine clauses that decide whether you get paid, and the mechanics of a work-made-for-hire clause — which often shares a sentence or two with the IP-infringement carve-out discussed above — are in the piece on work made for hire.
Frequently asked questions
What is the difference between an indemnification clause and a limitation of liability clause?
They move in opposite directions. Indemnification is about a third party — someone who isn't a party to your contract sues the client, or sues you, over something connected to the work, and the clause says which of you pays for that. A limitation of liability clause is about the two of you directly: it caps what either party can recover from the other for a breach of the contract itself, with no third party involved. A single dispute can touch both. If a client uses your design and a third party sues for trademark infringement, that is an indemnity question — who defends and pays the outside claim. If the client simply thinks your work was late or wrong and wants their fee back plus lost business, that is a liability question, and it is the liability cap, not the indemnity clause, that limits what they can collect from you for it. Contracts that discuss only one of the two have left half the risk undefined.
Is it normal for a freelance contract to have no cap on liability at all?
It is common in client-drafted paper, and it is not normal in the sense of being standard practice worth accepting without a look. A widely used reference agreement — the AIGA Standard Form of Agreement for Design Services — caps the designer's maximum liability at the total project fee and excludes indirect, incidental, and consequential damages outright (Section 10.4, aiga.org, checked 25 September 2026). That is the shape a fair cap takes: tied to what you were actually paid, not to the client's downstream losses. A contract that is silent on the cap does not mean there isn't one — it means the ordinary rules of contract damages apply instead, which in most US states allow recovery of losses that were reasonably foreseeable when the contract was signed, and foreseeable is a much larger number than your fee. Silence is not a favor to you.
My client says the certificate of insurance covers this, so the indemnity language doesn't matter. Is that true?
No, and it is worth separating the two documents in your head. The indemnity clause is a promise between you and the client about who pays if a claim comes in. The certificate of insurance is proof that a named insurer would pay a covered claim on your behalf, up to the policy limit, subject to the policy's own exclusions. A certificate does not enlarge or shrink what the contract says you owe — it is evidence of one possible funding source for it. If the indemnity clause is uncapped and your general liability policy caps out at one million dollars per occurrence, the gap between the policy limit and an actual judgment is still yours to cover personally. And a certificate naming the client as certificate holder is not the same as naming them an additional insured, which is the status that actually gives them a right to make a claim on your policy — the form itself says so (see the reading of ACORD 25 in the W-9 and insurance onboarding piece linked below).
What does 'duty to defend' mean if nobody has actually sued me yet?
It means the obligation to pay for a lawyer can start before anyone has decided who is at fault. Contracts that use the phrase 'defend, indemnify, and hold harmless' are usually creating two separate duties on different clocks. The duty to defend is typically triggered the moment a claim or lawsuit is filed that merely alleges something the indemnity clause covers — you don't wait to see who wins. The duty to indemnify is triggered later, once liability and damages are actually established, whether by judgment or settlement. Practice bulletins on construction contracts — where this distinction gets litigated often — describe the defend duty as broader and earlier-arising for exactly this reason (American Bar Association, Under Construction, checked 25 September 2026). If your contract has 'defend' in that sentence, read whether it also fixes who chooses counsel and who controls settlement, because whoever controls the defense usually also controls how much it costs.