Microsoft's AI Max for Search reaches general availability alongside a new account-linking security requirement, while Google tightens API access rules and rolls out fresh Demand Gen ad formats.
Microsoft has taken AI Max, its optional suite of Search campaign features designed to broaden intent matching and adapt to longer, richer queries, out of pilot and made it available to all advertisers.
Microsoft cites its own stats on Copilot engagement and purchase behaviour to justify the push, which I treat as vendor claims rather than fact. The underlying mechanic matters more than the headline number: this is Microsoft's version of broad match plus automated asset generation layered onto Search, and it will find you traffic you did not explicitly bid on.
New account-linking requests now require two identifiers instead of one, for both Manager account (CID) and Account (XID) linking. Platform users see this on new link requests from end of August with no technical changes needed. API users must update account-linking calls to include both identifiers, with enforcement running from 26 August through 22 September.
This is a genuine security tightening rather than noise, aimed at fraudulent account linking. The platform-side change is low friction, but anyone using the Microsoft Advertising API for account management, including agencies with scripts or in-house tooling, has a live enforcement window closing soon.
Updated Google Ads Developer Policies now require integrations to connect directly to Google Ads services through their own dedicated Google Cloud project, rather than routing through third-party proxies. Google says it is actively reviewing existing integrations and will contact developers who need to transition.
Proxies have been used by some reporting dashboards and bid management tools to simplify multi-account access. Google's stated rationale is cross-tenant data leak risk and DoS-related throttling, which is plausible given how proxy architectures work, though I cannot verify how widespread the actual incidents were.
Google is testing in-ad messaging conversations on YouTube Demand Gen ads, adding local activity and real-time offer personalisation for travel advertisers, and has made Multimodal Video Creation in Asset Studio generally available for building horizontal and vertical assets in one workflow.
Google credits prior Demand Gen changes with a 30 percent average lift in conversions or conversion value, a claim I treat with the usual scepticism given it comes from the platform grading its own product. The multimodal creative tool going GA is the more durable, testable change here.
On her Inside Google Ads podcast, Jyll Saskin Gales, a former Google employee and author of a book on audience targeting, describes what she calls non-linear targeting: building custom audience layers to reach ideal customers in industries where high CPCs or policy restrictions limit standard keyword or remarketing approaches.
Her core claim is that audience-based targeting can route around both cost and category restrictions that keyword-only strategies cannot solve. I agree this is a legitimate lever in genuinely restricted verticals such as some financial or health categories, but I would qualify it heavily: audience targeting without a hard revenue or offline conversion signal feeding the account is exactly the setup that lets automated bidding chase volume rather than value.