UK Publishers Push the CMA to Hold AI Assistants to Google's Attribution Bar

The UK's Competition and Markets Authority is finalizing a user choice conduct requirement for Google's general search services, with a decision expected by the end of 2026. In October, the Professional Publishers Association filed a response arguing AI assistants eligible for Google's choice screen must meet publisher safeguards equivalent to Google's own obligations.

Why choice screens matter for AEO

The CMA made eligibility criteria technology-neutral, meaning ChatGPT, Perplexity, and similar assistants could appear beside traditional search engines. That sounds pro-competition, but publishers note a structural difference: search historically sent clicks; AI answers often do not. AEO strategists must track which assistants gain default distribution on Android and Chrome in the UK.

PPA's equivalence argument

Google already faces publisher conduct requirements—controls, transparency, attribution—for AI features in search. The PPA wants any choice-screen AI that materially uses third-party content to match those standards before gaining placement. Without equivalence, the screen could steer users to services with weaker attribution than Google itself must provide.

Demands in detail

Recommendations include inline branded deep links, granular controls for indexing versus training versus agentic use, transparency reporting on impressions and clicks, and outcome-based assessments of referral traffic—not cosmetic badges alone. Publishers also ask Google to consult them when judging third-party eligibility.

Implications for brands

If rules tighten, answer engines may need licensed corpora or clearer citations, raising COGS for AI search startups. Brands with proprietary data may gain leverage in licensing talks. Marketers should document how each assistant cites their domain today to prepare for audits.

Global spillover

UK choice architecture influences EU DMA conversations and US antitrust remedies. Even US-only sites feel UK rules when assistants train on global crawls and when multinationals harmonize compliance.

Action list

Map your referral traffic by assistant where analytics allow. Join trade groups or file comments if you rely on organic discovery. Build direct audience channels before defaults reset user habits.

Summary

The CMA choice screen is not only about which icon users tap—it is about whether AI distribution rewards publishers fairly. AEO Soup readers should treat October 2026 comments as the blueprint for 2027 enforcement realities.

Across enterprise pilots, the pattern repeats: teams underestimate integration work and overestimate model magic. Budget two sprints for observability, one for access control, and ongoing red teaming before calling any October 2026 launch production-ready. Leaders who skip those steps publish demos, not products. Run tabletop exercises for credential leakage, model jailbreaks, and runaway agent loops before executives ask for a live customer pilot. Document rollback paths when a vendor changes terms, pricing, or model behavior without a semver notice.

Investors reading this week's headlines should separate infrastructure narratives from consumer hype cycles. Capex, containment, and compliance costs are real cash out the door; user counts without retention curves are marketing. Due diligence means asking vendors for incident postmortems, not only benchmark PDFs. Compare gross margin impact if inference becomes thirty or forty percent of COGS for software companies that previously shipped near-zero marginal cost. Ask whether moats are data, distribution, or merely early GPU reservations that expire when supply catches demand.

Journalists and creators covering these launches should link primary sources—company blogs, filings, and regulator dockets—so readers can verify claims as prices and policies shift. The pace of change in late October 2026 makes second-hand summaries stale within days. Readers deserve explicit dates on benchmarks, geographies on rollouts, and clarity when demos use cherry-picked prompts. When companies withhold model sizes or eval harnesses, treat performance claims as hypotheses until replicated.

Across enterprise pilots, the pattern repeats: teams underestimate integration work and overestimate model magic. Budget two sprints for observability, one for access control, and ongoing red teaming before calling any October 2026 launch production-ready. Leaders who skip those steps publish demos, not products. Run tabletop exercises for credential leakage, model jailbreaks, and runaway agent loops before executives ask for a live customer pilot. Document rollback paths when a vendor changes terms, pricing, or model behavior without a semver notice.

Investors reading this week's headlines should separate infrastructure narratives from consumer hype cycles. Capex, containment, and compliance costs are real cash out the door; user counts without retention curves are marketing. Due diligence means asking vendors for incident postmortems, not only benchmark PDFs. Compare gross margin impact if inference becomes thirty or forty percent of COGS for software companies that previously shipped near-zero marginal cost. Ask whether moats are data, distribution, or merely early GPU reservations that expire when supply catches demand.

Journalists and creators covering these launches should link primary sources—company blogs, filings, and regulator dockets—so readers can verify claims as prices and policies shift. The pace of change in late October 2026 makes second-hand summaries stale within days. Readers deserve explicit dates on benchmarks, geographies on rollouts, and clarity when demos use cherry-picked prompts. When companies withhold model sizes or eval harnesses, treat performance claims as hypotheses until replicated.

Across enterprise pilots, the pattern repeats: teams underestimate integration work and overestimate model magic. Budget two sprints for observability, one for access control, and ongoing red teaming before calling any October 2026 launch production-ready. Leaders who skip those steps publish demos, not products. Run tabletop exercises for credential leakage, model jailbreaks, and runaway agent loops before executives ask for a live customer pilot. Document rollback paths when a vendor changes terms, pricing, or model behavior without a semver notice.

Investors reading this week's headlines should separate infrastructure narratives from consumer hype cycles. Capex, containment, and compliance costs are real cash out the door; user counts without retention curves are marketing. Due diligence means asking vendors for incident postmortems, not only benchmark PDFs. Compare gross margin impact if inference becomes thirty or forty percent of COGS for software companies that previously shipped near-zero marginal cost. Ask whether moats are data, distribution, or merely early GPU reservations that expire when supply catches demand.

Journalists and creators covering these launches should link primary sources—company blogs, filings, and regulator dockets—so readers can verify claims as prices and policies shift. The pace of change in late October 2026 makes second-hand summaries stale within days. Readers deserve explicit dates on benchmarks, geographies on rollouts, and clarity when demos use cherry-picked prompts. When companies withhold model sizes or eval harnesses, treat performance claims as hypotheses until replicated.

Across enterprise pilots, the pattern repeats: teams underestimate integration work and overestimate model magic. Budget two sprints for observability, one for access control, and ongoing red teaming before calling any October 2026 launch production-ready. Leaders who skip those steps publish demos, not products. Run tabletop exercises for credential leakage, model jailbreaks, and runaway agent loops before executives ask for a live customer pilot. Document rollback paths when a vendor changes terms, pricing, or model behavior without a semver notice.

Investors reading this week's headlines should separate infrastructure narratives from consumer hype cycles. Capex, containment, and compliance costs are real cash out the door; user counts without retention curves are marketing. Due diligence means asking vendors for incident postmortems, not only benchmark PDFs. Compare gross margin impact if inference becomes thirty or forty percent of COGS for software companies that previously shipped near-zero marginal cost. Ask whether moats are data, distribution, or merely early GPU reservations that expire when supply catches demand.