How Sanders' AI Ownership Proposal Threatens AI-Powered Ads

If you have live spend in Performance Max, AI Max, Advantage+, or Advantage+ Creative Enhancements, the most important sentence in Bernie Sanders’ AI ownership proposal is not the one about a $7 trillion fund. It is the one that would require large AI companies to split their AI and non-AI businesses within 90 days of enactment.

The bill, introduced on June 18, 2026, says any company with more than $200 million in annual AI revenue would have to legally separate its AI operations from its non-AI operations, with no shared officers, directors, or equity ownership between the separated entities.[1] Read literally, that clause lands directly on the layer where Google and Meta now run modern ad delivery: AI bidding, audience modeling, creative assembly, measurement, and inventory allocation.

That does not mean Performance Max or Advantage+ is about to shut off. The bill is a long shot in the current Republican-controlled Congress, and there is no direct public response from Google or Meta in the cited material.[2] But it does mean the Bernie Sanders AI ownership proposal has a real ad tech impact to track, because the structural-separation language is aimed at the same integration layer that makes AI-powered campaign types work.

The clause that matters for ad platforms

The American AI Sovereign Wealth Fund Act would apply to companies that clear the $200 million annual AI revenue threshold. For a buyer, the operational problem is not just that Google and Meta would likely be covered. It is that their ad businesses are already built around AI systems that are not cleanly separable from the rest of the advertising machine.

Google’s AI operations, including Gemini, DeepMind, and Vertex, clearly put it above the bill’s AI-revenue threshold, while advertising contributes more than 80% of Alphabet’s $240 billion-plus annual revenue.[2] Meta’s AI operations, including LLaMA and Meta AI, also clear the threshold, while advertising represents 98% of revenue; Fortune also reported planned AI capital expenditure in the $125 billion to $145 billion annual range.[3]

The R Street Institute’s critique gets closer to the product issue than most political coverage. It says the separation clause would “sever the internal cross-subsidies” between AI and ad businesses and notes that “AI is baked into almost all their services now,” making separation unclear in practice.[4] Forbes reaches a similar legal-design objection, calling the separation requirement “impractical” and warning that “before long, every large company will be an AI company in some respect.”[5]

For paid media, that is the fault line. Performance Max is not a normal campaign type with a detachable AI module sitting politely next to it. Its value proposition depends on automated bidding, cross-inventory allocation, audience expansion, asset selection, and conversion learning. AI Max sits even closer to search-query interpretation and automated matching. Advantage+ and Advantage+ Creative Enhancements depend on Meta’s ability to use machine learning across delivery, creative variation, placement, and audience selection.

The exact impact on Performance Max, AI Max, Advantage+, and Advantage+ Creative Enhancements is not directly addressed by the bill text or by a reported Google or Meta statement. The product risk here is a reasoned architecture assessment: if the parent company must legally separate AI and non-AI operations with no shared officers, directors, or equity, the ad products sitting across that boundary become legally and commercially awkward to operate in their current form.

A legal split would not merely ask whether Google can keep Gemini in one subsidiary and ads in another. It would raise harder operating questions: who owns model outputs used in ad delivery, who controls performance data used for training or optimization, who sets product terms, who bears liability when an automated campaign changes creative, and whether the ad company can buy AI services from a separated AI company without recreating the integration the law was meant to break.

Why this is different from ordinary platform risk

Media buyers already live with opaque automation. A bidding model changes, an asset group behaves differently, a campaign starts pushing into inventory that looked marginal last month. Usually, that risk is internal to the platform: the system changes, and the buyer adjusts targets, budgets, exclusions, feeds, or creative inputs.

The Sanders clause is different because it targets ownership and control, not a feature toggle. A normal product change can be rolled out behind the interface. A structural-separation rule would ask whether the company that sells the ad product is legally allowed to share governance, equity, and operating control with the AI system that makes the ad product perform.

That matters because AI-powered ad buying is already material. IAB reports that AI-powered ad buying accounts for about 8% of U.S. ad revenue, or roughly $35 billion, and is scaling quickly.[6] Adoption does not prove effectiveness, and it does not tell us how much of that spend would be impaired by one bill. It does show that the affected surface is no longer experimental.

The practical exposure is highest where buyers have let campaign structure collapse into platform automation. A cleanly segmented search account can still be managed around keywords, match types, bids, and landing pages. A Performance Max or Advantage+ account has fewer buyer-controlled levers. If the platform has to re-architect the AI layer underneath it, the buyer has fewer places to hold the system steady.

Three operating scenarios to monitor

The bill’s current odds should keep this out of the emergency channel. They should not keep it out of the tracker. The useful exercise is not predicting passage as if it were imminent; it is knowing what signals would matter if the idea moves from political proposal into legislative bargaining, litigation, or agency rulemaking.

Full compliance: the disruptive version

Full compliance is the scenario buyers should understand first, even if it is not the most likely near-term outcome. Under the bill’s text, covered companies would have 90 days after enactment to separate AI and non-AI operations, with no shared officers, directors, or equity.[1] That is a hostile timeline for products whose AI and ad-delivery systems have been co-developed for years.

In that version, Google and Meta would have several bad choices. They could pause or limit certain AI-dependent campaign features while they rebuild governance and data flows. They could spin AI services into a separate entity and license those services back to the ad business. They could keep the product names but narrow what the automation is allowed to do. Or they could absorb new compliance, contracting, and margin pressure that eventually changes pricing or product packaging.

For buyers, the risk would show up less like a single account outage and more like a sequence of product constraints: new terms around automated assets, changed data-use language, restrictions on measurement signals, renamed AI features, altered campaign eligibility, or interface warnings that separate “AI-generated” functions from standard delivery. The uncomfortable part is that a campaign might remain technically live while the optimization logic underneath it changes.

A re-architected Performance Max could still exist, but it might behave differently if model training, creative generation, conversion prediction, and ad serving must move through arm’s-length contracts. A re-architected Advantage+ could still buy impressions, but its strongest advantage comes from Meta’s ability to combine delivery signals, creative feedback, user behavior, and advertiser outcomes inside one system. That is exactly the kind of integration the clause puts under pressure.

Litigation: uncertainty without immediate account change

The more realistic near-term disruption, if a similar proposal ever advanced, would be legal uncertainty. Google and Meta would be expected to challenge a forced separation of AI and non-AI operations. The cited material does not include direct platform responses to this bill, so any platform strategy here is inferred from the stakes, not reported as a stated position.

A litigation track would probably leave campaigns running while legal teams, trade groups, and regulators argue over definitions: What counts as AI revenue? Is an ad-ranking model an AI business or an ad-business tool? Can a separated AI entity provide services to an affiliated ad company if equity and governance are split? Does measurement data become an AI asset, an advertising asset, or both?

That is not an immediate media-plan failure. It is a planning problem. The worst accounts to manage during that kind of uncertainty are the ones with no fallback structure: no standard shopping or search coverage, no independent creative testing rhythm, no source-of-truth measurement outside the platform, and no budget model for reduced automation.

A narrow ad-tech exemption: plausible, not bankable

The exemption path is easy to imagine. Lawmakers could decide the clause should target frontier model labs, cloud-scale AI infrastructure, or general-purpose AI systems, while carving out routine AI used inside advertising, search ranking, recommendation, fraud detection, or measurement. That would make the bill less disruptive to existing commercial software.

But an exemption would need careful drafting. If ad-tech AI is exempt because it is “just optimization,” the exemption could swallow most platform AI. If it is not exempt, then the law collides with the operating reality that the largest ad platforms are already AI companies in the sense Forbes warned about.[5] A buyer should not assume the safe version wins before seeing actual exemption language.

Do not confuse the stock tax with the product threat

The stock-transfer mechanism has pulled much of the coverage because it is easier to headline. Tax Notes, via Thomson Reuters, describes a one-time 50% stock transfer mechanism valued at about $7 trillion at current market capitalizations, and notes financial pressure around AI firms including OpenAI, which lost 237% of revenue in 2024, and Anthropic, which was not yet profitable.[7]

That could matter for ad buyers if financial pressure eventually changed cloud pricing, platform margins, or the cost of AI-heavy ad features. But the pass-through path is indirect and contested. A one-time equity transfer is not the same thing as a per-impression tax on Performance Max or Advantage+.

The structural-separation clause is more direct because it asks whether the AI system and the advertising business can remain under the same operating roof at all. If the answer becomes no, the buyer is not just paying for the same product under a different cost base. The product may have to be rebuilt.

What buyers should watch next

The correct posture is neither panic nor dismissal. The bill is unlikely to pass as written, but it creates a dated regulatory marker: June 18, 2026, the day a federal proposal put forced AI/non-AI separation on the table for companies whose ad products depend on integrated AI.

• Track bill movement: committee activity, companion legislation, amended definitions of “AI revenue,” and any carve-outs for advertising, search, cloud, recommendation systems, or measurement.
• Watch Google and Meta statements: especially comments in earnings calls, public-policy blogs, trade-association filings, and litigation posture if similar language gains traction.
• Monitor product documentation: changes to Performance Max, AI Max, Advantage+, and Advantage+ Creative Enhancements docs may surface before interface changes do.
• Read terms and data-use updates: separation risk will likely show up in language around automated decisioning, model improvement, creative generation, measurement signals, and third-party service providers.
• Keep fallback campaign structure alive: not because collapse is imminent, but because accounts fully dependent on black-box automation have the least room to adapt if product rules change.

This belongs beside other AI advertising regulation entries, including Tesla’s Full Self-Driving ruling and AI advertising regulation. That case is about false-advertising enforcement, not platform separation, but both point in the same direction: regulators are moving from vague AI concern toward specific operating constraints.

The bounded judgment is simple. This is not imminent collapse for AI-powered ads. It is also not political theater that media buyers can safely ignore. The separation clause is a direct regulatory signal aimed at the integrated AI-plus-ads architecture that Google and Meta have spent years making harder for advertisers to inspect and harder to replace.

References

1. NEWS: Sanders Introduces Legislation to Create $7 Trillion AI Sovereign Wealth Fund, U.S. Senator Bernie Sanders, June 18, 2026.
2. Sovereign wealth fund, tax on AI companies unveiled by Sanders, Roll Call, June 18, 2026.
3. Bernie Sanders AI ownership sovereign wealth fund electrification, Fortune, June 3, 2026.
4. The Sanders AI Sovereign Wealth Fund Act Is a Death Sentence for American Technology Leadership, R Street Institute.
5. Bernie Sanders Wants a U.S. Sovereign Wealth Fund for AI, Forbes, June 22, 2026.
6. AI Adoption Is Surging in Advertising, But Is the Industry Prepared for Responsible AI?, IAB.
7. Sanders Calls for Tax on Systemically Important AI Activity Payable in Equity, Tax Notes / Thomson Reuters.