Google Ads New Terms of Service — 1 July 2026: What Every UK SME Must Do Now That AI Automation Owns Your Ad Spend

On 1 July 2026, a quiet clause change in Google’s Advertising Programme Terms rewired the relationship between every UK advertiser and the automation that now runs their campaigns. The updated terms formally extend Google’s right to use advertiser inputs — landing-page URLs, conversational prompts typed into Gemini, and the content of your own website — to generate ad assets across the full suite of AI-powered campaign features. In the same breath, the terms place a “continued obligation” on the advertiser to review every automatically generated asset for accuracy, policy compliance and ownership. In plain English: Google’s AI writes the ad, and you are legally responsible for what it says.

For UK SMEs this is not an abstract legal footnote. It is a direct transfer of accountability onto the exact businesses least equipped to carry it — the ones running Google Ads without a dedicated in-house PPC team, relying on Performance Max, automatically created assets and Gemini-generated headlines to keep campaigns live. With 70 million AI-generated ad assets served in Q4 2025 alone, the use of Gemini-generated assets up threefold across 2025, and UK Consent Mode having become the sole gatekeeper for advertising data since 15 June 2026, the compliance and financial exposure sitting inside an unmanaged Google Ads account has never been greater. This briefing breaks down what actually changed on 1 July, why it matters specifically to UK small and medium-sized businesses, and where the human oversight layer now has to live.

What actually changed on 1 July 2026

The 1 July update to Google’s Advertising Programme Terms does two things at once. First, it broadens and makes explicit the licence you grant Google over your inputs. When you point a campaign at a landing-page URL, type a description of your business into Gemini’s conversational campaign builder, or let Google crawl your website to build automatically created assets, you are now formally authorising Google to use that material to generate headlines, descriptions, images and other creative across its AI features. This is the codification of a machine that has been running since early 2024 — Gemini conversational campaign creation arrived for UK advertisers in January 2024, and automatically created assets rolled out in February 2024. The terms have finally caught up with the technology.

Second, and far more consequentially for the account holder, the update attaches a “continued obligation” to that grant. Google’s position is unambiguous: the automation may generate the asset, but the advertiser retains the ongoing responsibility to review what it produces and ensure it complies with Google’s policies and applicable law. If Performance Max assembles a headline advertising a price you no longer offer, if AI Max for Search stitches together a claim you cannot substantiate, or if an automatically created asset pulls an image or a phrase you do not hold the rights to, the liability for that ad content sits with the account holder — not with the model that wrote it.

This matters because the direction of the product has been to remove humans from the per-asset approval loop entirely. Performance Max campaigns generate and rotate creative combinations automatically. AI Max for Search — the successor to Dynamic Search Ads, which Google confirmed will be retired and upgraded to AI Max in September 2026 — extends that same generative approach to the search network. AI Max for Shopping launched on 30 April 2026 with AI-generated text customisation drawn directly from Merchant Center feeds. The combined effect is that the majority of ad content in a modern Google Ads account is now generated without a person signing off on each individual line before it serves. The 1 July terms confirm that the absence of that sign-off does not reduce the advertiser’s responsibility — it simply moves the review obligation to after the asset is live, where most SMEs never look.

How the automation-accountability shift built up

The 1 July 2026 terms did not appear from nowhere. They are the formal endpoint of a two-and-a-half-year march from optional automation to default automation, running in parallel with a broader tightening of the rules around advertising data. The timeline below traces how UK advertisers arrived at a point where AI owns the ad and the SME owns the risk.

Where the exposure concentrates across a Google Ads account

Not every part of a Google Ads account carries the same accountability risk under the 1 July terms. The exposure concentrates in the features that generate creative automatically and serve it without a per-asset human sign-off. The bar chart below ranks the major campaign surfaces by how much unreviewed, machine-generated content they typically put in front of UK audiences — and therefore how much of the “continued obligation” each one loads onto the account holder.

The pattern is clear: the newer and more automated the format, the more unreviewed content it serves and the more of the review obligation it transfers to the advertiser. Performance Max sits at the top because it combines every asset type — headlines, descriptions, images, video, product data — into machine-selected combinations that no human approves individually. When Dynamic Search Ads becomes AI Max for Search in September 2026, the search network moves decisively up this chart. A UK SME that has “set and forgotten” a Performance Max campaign is, under the 1 July terms, running the single highest-exposure surface in the entire platform without the review cadence the terms now assume.

How many UK SMEs run Google Ads without a review layer?

The core problem is structural. UK SMEs overwhelmingly run Google Ads without dedicated in-house PPC staff — a marketing manager, an owner, or an office administrator manages the account alongside a dozen other responsibilities. That is precisely the profile Google’s automation was designed to serve, and precisely the profile least able to meet a “continued obligation” to review every automatically generated asset. The donut below reflects a best estimate of the share of UK SME Google Ads accounts that have no formal, recurring asset-review process against the 1 July requirements.

The precise figure is unknowable — Google does not publish it — but the logic is robust. When automation is the default, when Performance Max and AI Max are actively promoted as the low-effort route to results, and when the typical UK SME account is managed by someone without a PPC title, the natural outcome is that most accounts have no weekly cadence for opening the assets report, reading each machine-written headline, and checking it against current pricing, current claims and current rights. The 1 July terms do not create this gap. They convert a pre-existing operational gap into a contractual and, potentially, a legal one — and they do it at the moment AI is generating more ad content than at any point in the platform’s history.

Where UK SMEs are most exposed under the new terms

The score card below breaks the accountability risk into eight concrete factors. Each is a point at which the 1 July terms, the 15 June Consent Mode change, or the underlying automation creates exposure that a UK SME running Google Ads without a managed partner is likely carrying right now. High-rated rows are the ones that combine unreviewed machine content with direct legal or financial consequence.

Any UK SME scoring High on the first four rows is carrying real exposure today. The combination of an unreviewed generative campaign surface and a misconfigured Consent Mode means the account is simultaneously serving content the business has not checked and potentially handling advertising data without a clean lawful basis. Neither of these is visible from the account’s headline performance metrics — a campaign can be delivering a healthy return on ad spend while quietly running a policy-violating headline and a broken consent signal underneath.

What the exposure costs by business size

The cost of the accountability shift is not a single number — it scales with ad spend, the number of live automated campaigns, and the sensitivity of the sector. The table below sets out indicative exposure bands for UK organisations of different sizes, covering the realistic consequences of running generative Google Ads campaigns without a review layer: wasted spend on off-target AI creative, the cost of a policy suspension, and the management effort the “continued obligation” now implies.

These bands do not include the hardest-to-quantify cost: a Google Ads account suspension for a repeated policy violation. When an automatically generated asset trips a policy — for example by making an unsubstantiated claim, using a restricted term, or referencing content the advertiser does not own — the enforcement action lands on the account, not on the automation. For a UK SME that depends on Google Ads for lead flow, a suspension is not a line item; it is an interruption to revenue that can take days or weeks to resolve. The “continued obligation” clause is, in effect, Google telling advertisers that preventing this is now explicitly their job.

Automation alone versus automation with human oversight

The distinction is not “automation versus no automation”. The automation is not the problem, and turning it off is neither practical nor advisable — Performance Max and AI Max genuinely deliver reach and efficiency that manual campaigns cannot match. The distinction is whether there is a human oversight layer sitting on top of the automation, doing the specific work the 1 July terms now require: reviewing the generated assets, controlling the inputs the AI draws from, and keeping the measurement plumbing compliant. That layer is exactly what a managed Google Ads partner provides and exactly what an unmanaged SME account lacks.

At-a-glance: the 1 July 2026 Google Ads changes

How this connects to the wider 2026 picture

The Google Ads accountability shift is one thread in a year defined by AI moving from tool to autonomous actor — and by the responsibility for what that AI does landing on the businesses that deploy it. The same pattern runs through our recent coverage. The rise of autonomous AI in offensive security is examined in our briefing on JadePuffer, the first agentic AI-driven ransomware, where an LLM makes attack decisions without a human in the loop — the mirror image of an LLM making advertising decisions without one. The governance gap that leaves UK SMEs exposed to both is the theme of our analysis of the Superscript cyber report showing 51% of UK SMEs breached, which traces so many incidents back to the absence of a proactive oversight layer. And the board-level case for treating this kind of exposure as a strategic priority — rather than an operational afterthought — is set out in our piece on Barclays’ Q1 2026 UK cyber investment data and the Virtual CIO response. For SMEs weighing how much of their operations to hand to automation, our look at cloud misconfiguration in the 2026 DBIR makes the same underlying point: automated systems fail quietly, and the cost of not looking is paid by the account holder.

Frequently asked questions

CloudSwitched

London-based managed IT services provider offering support, cloud solutions and cybersecurity for SMEs.