- Since July 1, 2026, Google’s advertising terms say you authorize Google to use “automated Program features to format, select, or generate Targets, Ads, or Destinations” on your behalf.
- That clause hands no liability to Google. The next sentence keeps you “solely responsible” for whatever those features produce.
- It is also not the clause that can get your account suspended. A separate restriction bars you from authorizing any third party to scrape or extract Google ad data.
- Microsoft’s agreement uses nearly the same sentence. It authorizes automated tools to “format ads” only, and keeps selection and generation opt-in.
- The audit worth running is not “am I using automation.” It is “does each tool in my stack reach Google through the API or through a browser.”
Most of the coverage of Google’s July terms update asks whether the machine writes good ads. That is a media buying question, and we graded generated copy on cost per qualified call separately.
This is the other question. You are a party to a contract that changed. Something was authorized on your behalf, and your account carries the consequences.
So I pulled the actual terms document Google publishes, not an agency recap of it. Then I pulled Microsoft’s equivalent agreement to see whether the sentence is standard boilerplate or a real shift. It is a real shift, and the difference is two verbs.
Quick answers:
- Who is liable for ads Google generates automatically?
- Can you opt out of Google Ads automated ad generation?
- Do third-party Google Ads tools violate the terms of service?
- What gets a Google Ads account suspended without warning?
- When did the new Google Ads terms take effect?
What the clause actually says
Here is the sentence, from the Google LLC Advertising Program Terms that Google itself posts:
“The Program is an advertising platform on which Customer authorizes Google and its affiliates to serve Ads, including through the use of automated Program features to format, select, or generate Targets, Ads, or Destinations on Customer’s behalf.”
Read the next sentence too, because it is the one that decides who pays:
“Customer will continue to be solely responsible for all Targets, Ads, or Destinations arising out of Customer’s use or continued use of such Program features, and for reviewing and, as applicable, approving or removing relevant campaigns and assets, including for compliance with Policies.”
That pairing is the whole structure. Google took a permission. You kept the liability.
A note on what I could and could not confirm. The clause above is verbatim from Google’s own posted document, which carries the date July 1, 2026 on its final line. The prior wording is a different story. Google publishes only the current version. Archived copies of that document endpoint do not return readable text. Search Engine Land reported the change on June 2, 2026, ahead of the rollout. It characterized the older language as letting advertisers opt in or out of many automation features. I could not verify that older text against a primary source, so I am not going to quote it. What is verifiable is the current wording, and what it does.
What “format, select, or generate” covers in your account
The three nouns in that clause are defined terms, and the definitions are broader than most summaries let on. Section 1 of the same document defines them.
“Targets” are your “Ads trafficking or targeting decisions (e.g., keywords).” That is your keyword and audience layer.
“Destinations” are the places your ads send people. That definition runs long. It covers “domains, landing pages, mobile applications” plus “the related URLs, accounts, waypoints, and redirects.”
“Ads” is defined together with “Creative” and covers your “advertising materials, inputs, feed data, and technology.”
Put those back into the sentence and the scope gets concrete. The authorization is not limited to headline text. It reaches keywords, landing page selection, redirects, and feed data.
One more line from the same section sets the ceiling on any argument you might want to have about it later:
“Google and its affiliates or Partners may reject or remove a specific Target, Ad, or Destination at any time for any or no reason.”
The practical takeaway for a lead-gen operator is about destinations. If automated features can select or generate a destination, then your landing page set is inside the automated surface, not outside it. That matters more than the ad copy does, because the destination is where your form, your consent language, and your call routing live.
Microsoft wrote the same sentence, and left out two verbs
This is the part no one seems to have checked, and it is the fastest way to see what actually changed.
Microsoft’s advertising agreement contains a sentence built on the identical frame. Here it is, from the Microsoft Advertising Agreement:
“Microsoft Advertising is an advertising platform on which Company authorizes Microsoft and its partners to use automated tools to format ads.”
Same construction. Different verb list. Microsoft’s baseline authorization covers formatting only.
Selection and generation live in the next sentences, and they are handled as an opt-in:
“Microsoft and its partners may also make available to Company certain optional features … to assist Company with the selection or generation of ads. Company is not required to authorize use of these optional features and, as applicable, may opt-in or opt-out of using these features. However, if you use these features, then you will be solely responsible for the ads.”
Line the two up and the diff is clean. Both platforms make you responsible for the output. Microsoft keeps “selection or generation” as optional features you can decline in the contract. Google folded format, select, and generate into one baseline authorization, and attached them to targets and destinations as well as ads.
That is the change worth telling a client about. Not “Google uses AI now.” Everyone knew that. The contract stopped describing generation as something you elect.
Nobody clicked accept, and that was allowed
There was no consent screen, and the terms explain why none was needed.
The change-of-terms section says material changes get advance notice. Revised terms are posted at google.com/ads/terms. Changes “will become effective 7 days after posting.” Then it closes the loop:
“Continued use or access to the Programs following the posting of changes to these Terms constitutes Customer’s acceptance of any such changes.”
Running a campaign on July 2 was the acceptance. That is ordinary for platform contracts, and it is why the effective date is the thing to diary rather than an email you are waiting for.
If you manage accounts for clients, this is worth a plain sentence in your next report. Your client is the contracting party. They accepted revised terms by continuing to spend, and most of them do not know it happened.
The clause that can actually suspend you is a different one
Here is where the popular framing goes wrong. The authorization clause is not an enforcement clause. It grants Google permission. It does not describe anything you can violate.
The clause that creates real account exposure sits elsewhere in the same document, and it is aimed squarely at third parties:
“Customer will not, and will not authorize any third party to, (i) generate automated, fraudulent or otherwise invalid impressions, inquiries, clicks or conversions … (iii) use any automated means or form of scraping or data extraction to access, query or otherwise collect Google advertising-related information from any Property except as expressly permitted by Google, or (iv) attempt to interfere with the functioning of or misuse the Programs.”
Read part (iii) slowly. You are responsible not only for what you do, but for what you authorize a third party to do. Every tool you granted access to is a third party you authorized.
The policy side carries the teeth. Google’s Abusing the ad network policy includes a Circumventing systems provision:
“Engaging in practices that circumvent or interfere with Google’s advertising systems and processes, or attempting to do so is not allowed.”
And the enforcement language on that page is not graduated:
“If violations of this policy are found, your Google Ads accounts will be suspended upon detection and without prior warning, and you will not be allowed to advertise with Google Ads again.”
The same page notes these violations are “considered egregious.” Accounts “are only reinstated in compelling circumstances.” There is no warning email in that sequence. The appeal path also got narrower this month. Google closed in-account appeals at six months on July 21 and capped appeals at three per ad.
So the risk is real. It just has nothing to do with the clause going around.
Which third-party tools are actually exposed
The question worth asking is which tool categories now carry flag or suspension risk. Having read the actual clauses, the honest answer is that the July change did not create new risk for any of them. The exposure was already there, and it splits on one line: API or browser.
Google runs a sanctioned path for third-party tools. The Google Ads API policies set out what a compliant tool must do. The enforcement ladder there is ordinary commercial stuff. Google contacts the provider by registered email with a chance to correct. Then come non-compliance fees, then a downgrade from Standard to Basic access, then token termination. That is a vendor problem with a notice period. It is not the no-warning suspension path.
Tools that reach Google by driving a logged-in browser session are in a different position, because scraping and data extraction are what part (iii) names.
Run your stack against that line:
| Tool category | Usual access path | Where the risk sits |
|---|---|---|
| Bid management platforms | Google Ads API, developer token | Vendor-side compliance, notice-based enforcement |
| Automated creative and asset tools | API, or asset upload | Policy compliance of the output, not the automation |
| Google Ads scripts | First-party, runs inside your account | Sanctioned surface; risk is what the script does |
| Feed management tools | API or Merchant Center | Data accuracy and destination policy |
| Rank and competitor scrapers | Headless browser against Google properties | Part (iii), scraping and data extraction |
Two clarifications that keep operators out of trouble.
Google Ads scripts are a first-party feature. Running scripts is not third-party access, and the July change did not touch them. What can hurt you is a script that does something the policies prohibit, which is a content question rather than an access question.
Bid management platforms using a developer token are doing exactly what the API exists for. Using automation to manage bids has never been the violation. If anything, the bid strategy layer is where Google most wants you automated.
The one to actually look at is the tool nobody remembers buying. Competitive intelligence dashboards and rank trackers frequently pull data by scraping search results. That data is genuinely useful, which is why it lives in a lot of agency stacks unexamined.
The stack audit you can run this week
This takes an afternoon and produces a document you can hand a client.
One. List every entity with access to your Google Ads accounts. Pull it from Tools, then Access and security, and separately check Linked accounts. Include the vendors nobody logs into anymore.
Two. For each one, answer a single question: does it authenticate through the Google Ads API with a developer token, or does it log into a browser session? Vendors answer this readily, and a vendor that will not answer it in writing has told you something.
Three. Separate the tools that manage your account from the tools that collect data about Google. Management through the API is the sanctioned path. Collection by scraping is what the restriction names.
Four. Check whether any reporting tool delays your Google Ads data by more than 24 hours. The API policies require a provider to “prominently disclose this delay.” An undisclosed delay is a signal about how the vendor treats the rest of the rules.
Five. Confirm nobody is asking your clients to apply for their own API token to use a tool. The API policies prohibit requiring that, and it is a common sign of a tool operating outside the intended model.
Six. Revoke what you are not using. Dormant access is the cheapest thing on this list to fix and the easiest to forget. If your toolchain touches the API directly, the v25 release also forces code changes this quarter. The audit and the migration can share a calendar slot.
What to leave on, what to turn off, what to put in writing
Leave the bidding automation on. That is not where this risk lives, and turning it off to feel safer costs performance for nothing.
The contract authorization has no opt-out, but that is a different thing from having no controls. Feature-level settings still exist in the product, and this is the distinction most summaries blur. You cannot decline the clause. You can still decide which automated features run, and you should decide deliberately rather than by default, especially on destinations.
Turn a hard eye on anything that generates or selects destinations. Your landing page is where consent language and call routing live. In regulated verticals, an automatically selected destination is a compliance surface, not a creative one. If you buy calls in health or finance, pair that review with the Limited Ad Serving policy, because the two interact.
Retire or replace anything that scrapes. Not because of the July change, but because part (iii) has been sitting there and the enforcement language is unusually blunt.
Read the verbs, not the headline
The useful skill here is smaller and more durable than this one update.
Platform terms change by verb list. Microsoft’s sentence and Google’s sentence are structurally identical and differ by two words, and those two words are the entire story. No summary I read caught it, because summaries paraphrase and paraphrase is where verbs go to die.
The next revision will arrive the same way. Posted quietly, effective seven days later, accepted by the fact that you kept spending. When it does, open both documents and diff the verbs. That takes fifteen minutes and it is the only version of this that does not depend on someone else reading carefully for you.
Frequently Asked Questions
Who is liable for ads Google generates automatically?
You are. Google’s terms authorize automated features to format, select, or generate targets, ads, and destinations. The next sentence states that the customer “will continue to be solely responsible” for all of them. That responsibility explicitly includes “reviewing and, as applicable, approving or removing relevant campaigns and assets, including for compliance with Policies.” The authorization widened. The liability did not move.
Can you opt out of Google Ads automated ad generation?
Not at the contract level. The authorization in the Advertising Program Terms carries no opt-out, and continuing to run campaigns constitutes acceptance. Feature-level controls inside the product are a separate matter and still exist, so you can still choose which automated features run in a given account. Microsoft takes the other approach and keeps selection and generation as optional features an advertiser may opt in or out of.
Do third-party Google Ads tools violate the terms of service?
Most do not. Tools that connect through the Google Ads API with a developer token are using the path Google built for them. Enforcement there runs through the provider, with notice, fees, and access downgrades. The exposure sits with tools that use “automated means or form of scraping or data extraction” against Google properties. The terms bar you from authorizing a third party to do that.
What gets a Google Ads account suspended without warning?
Google’s Abusing the ad network policy treats circumventing or interfering with its advertising systems as an egregious violation. The policy states that accounts “will be suspended upon detection and without prior warning,” and that reinstatement happens only “in compelling circumstances.” That is the enforcement path worth designing your stack around, and it is separate from the July automation clause.
When did the new Google Ads terms take effect?
The Google LLC Advertising Program Terms currently posted carry the date July 1, 2026. Under the change-of-terms section, revised terms become effective seven days after posting, and continued use of the programs constitutes acceptance. No advertiser was asked to re-accept anything.
