Introduction
Since this guide was first published in 2023, the use of AI in advertising has moved from experimentation to mainstream adoption. The generative AI advertising market continues to grow exponentially into a multibillion-dollar industry. Alongside this growth, regulators have moved aggressively to establish guardrails, new advertising formats have emerged within AI chatbot conversations, and advertisers are confronting novel risks – from platform AI agents modifying ads without permission to AI-generated imagery that amplifies harmful stereotypes.
Regulatory developments: Enforcement intensifies
The Federal Trade Commission (FTC) has dramatically increased its focus on AI-related advertising practices, concentrating enforcement on two areas: companies overhyping their AI capabilities (often called “AI washing”) and undisclosed AI-generated endorsements and reviews. AI-washing cases have targeted businesses making inflated or unsupported claims about their AI technology’s capabilities – such as marketing services as powered by AI when they were not – and, as of early 2026, the FTC has brought more than a dozen actions, including multimillion-dollar settlements.
On the endorsement side, the FTC effectively requires a “double disclosure”: campaigns involving both a paid relationship and AI-generated endorsements must disclose both facts conspicuously (for additional considerations around the use of synthetic performers in advertising, see Talent and Use of Synthetic Performers. This requirement flows from two overlapping sources: (1) the FTC’s 2023 Endorsement Guides, which cover advertising messages that consumers are likely to believe reflect the opinions or beliefs of someone other than the sponsoring advertiser; and (2) the Rule on the Use of Consumer Reviews and Testimonials, which prohibits misrepresenting a reviewer’s identity, experience, or existence.
Political dynamics have influenced enforcement priorities. In December 2025, the FTC vacated its consent order against an AI-powered writing assistant that could generate testimonials and reviews, signaling a shift toward requiring concrete – not speculative – consumer harm. The Commission extended this posture in July 2026 with a proposed policy statement on AI “accuracy,” taking the position that secretly steering AI outputs toward undisclosed objectives – including to comply with state algorithmic-discrimination laws – could constitute deception under Section 5. For advertisers, this means claims that AI tools are “accurate” or “unbiased” will be measured against actual system behavior, and third-party AI vendors should be vetted for covert output tuning. Consistent with a December 2025 executive order, the Commission has also taken the position that such state laws may be impliedly preempted to the extent they conflict with federal regulatory schemes. Whether state AI laws will ultimately be preempted remains uncertain; for now, advertisers should monitor both state and federal developments.
Market shifts: AI chatbot advertising arrives
A significant market development is the arrival of advertising within AI chatbot conversations. This format is fundamentally different from traditional digital advertising because it uses contextual targeting (no cookies, no behavioral profiles) and raises a novel compliance challenge: consumers perceive AI chatbot responses as dialogue or advice, not as a search results page with labeled ads. In particular, AI chatbot operators will need to navigate the patchwork of advertising disclosure laws to ensure consumers are well-informed about what is being communicated to them through these chatbot conversations. Many states have enacted chatbot-specific laws, and the FTC has signaled this area will face heightened scrutiny at the federal level.
Many brands have deployed their own AI chatbots for shopping assistance, customer service, and lead generation – raising an important disclosure question: does the consumer know they are talking to a bot, and is the bot’s “recommendation” actually a paid placement or a self-interested placement? California’s BOT Disclosure Act (Bus. & Prof. Code Section 17940) already requires clear disclosure when a bot communicates with a California resident to influence a purchase, with penalties of up to $2,500 per violation, and similar reasoning could extend to branded shopping assistants that steer consumers toward a company’s own products.
A broader wave of state “companion chatbot” laws, together with the EU AI Act’s Article 50, also require operators to disclose non-human status and, in several states, to provide safety protocols for minors. While most of these laws target chatbots designed to simulate a personal relationship rather than commercial chatbots, some, such as California’s, carry a private right of action, underscoring a fast-moving legislative trend toward mandatory AI disclosure that advertisers deploying consumer-facing chatbots should monitor. At the federal level, the FTC opened a 2025 inquiry into several major AI companies regarding the effects of companion chatbots on children, though, as of the time of writing, the agency has not yet brought a chatbot-specific enforcement action.
Regulators themselves are also harnessing AI. In 2025, the FDA deployed AI tools to proactively scan television, print, and digital channels for non-compliant pharmaceutical advertising. This marks a shift from reactive, complaint-driven enforcement to proactive, AI-surveillance-based monitoring.
Rogue AI agents: When platforms create ads without permission
Major ad platforms’ AI tools can now autonomously modify or create advertising content without advertiser approval. In one reported instance, an agency discovered AI-generated music on a client’s video ad – a soundtrack no one had chosen, added automatically via a single campaign-settings toggle.
This autonomous creative substitution is a form of agentic AI – systems that independently plan and execute multi-step tasks with minimal human intervention. Major ad platforms have adopted agentic tools that can select creative, keywords, and placements with limited advertiser visibility, execute entire campaigns from minimal inputs, or generate complete video ads from a text prompt. Because these tools generate and place content with little or no human sign-off, they meaningfully increase intellectual property (IP) and brand safety risks, particularly as platform terms typically limit platform liability.
AI creative failures also raise false advertising exposure. In one case, an outdoor retailer was auto-enrolled in a feature that produced a social ad (falsely) showing a bicycle with two handlebars. In another, a platform’s automation pulled a link for pet vaccinations into an ad meant for cattle vaccines. Where AI-generated content includes false information about a product or service, advertisers can face liability under Section 5 of the FTC Act and the Lanham Act for misrepresenting the goods or services offered. Some platform policies are beginning to police this, for example by only permitting AI-assisted edits such as background replacement, while prohibiting AI generation that misrepresents a product’s physical characteristics. However, because platform settings can quietly revert or introduce new automated options, advertisers should conduct regular manual audits of campaign settings.
AI bias and stereotyping: A growing brand risk
AI-generated advertising images have sparked public backlash when they perpetuate racial or gender stereotypes. In one incident, a major brand pulled an AI-generated ad campaign after content creators condemned it for amplifying harmful stereotypes – demonstrating how AI tools can reproduce and even intensify biases present in their training data. Even campaigns that use AI ironically or critically can cause real harm if the AI-generated content reinforces existing stereotypes. The risk extends beyond intent to the visual output itself, as stereotyped imagery may lead to negative PR, private lawsuits from impacted individuals, and regulatory scrutiny. This emphasizes the importance of diverse human review of AI-generated content and imagery, particularly when depicting people, rather than relying on automated content moderation alone.
AI-driven bias has also attracted regulatory attention, although state and federal regulators have taken different approaches. On August 7, 2026, the FTC issued a formal policy statement announcing it will no longer pursue claims based on disparate-impact or “unfair discrimination” theories under Section 5 of the FTC Act or any other statute that it enforces. As a result, the risk of enforcement is primarily driven by state attorneys general. For example, the Massachusetts AG settled with a student loan lender over allegations that its AI algorithmic underwriting resulted in disparate impacts on minority applicants. As part of the settlement, the AG required the company to create an internal algorithmic oversight team and develop policies addressing algorithmic bias and accountability. Given that a bipartisan coalition of state AGs, governors, and legislators successfully opposed a federal provision in the “Big, Beautiful Bill” that would have preempted state AI laws, the states remain active enforcers, relying on Unfair and Deceptive Acts and Practices (UDAP) statutes, state consumer protection laws, and civil rights and anti-discrimination statutes.
Practical guidance for advertisers
- Document AI usage in creative workflows. Records of AI tool usage, human oversight, and guardrails are now legal assets, not merely internal process documents.
- Embed disclosure logic into content systems. For multi-state campaigns, build disclosure compliance into content creation workflows by default, rather than applying it as an afterthought.
- Update influencer agreements. Add specific parameters regarding AI use and synthetic content generation to all influencer contracts.
- Monitor chatbot disclosure obligations. Confirm whether branded chatbots trigger state bot-disclosure or AI-transparency laws in addition to FTC “clear and conspicuous” standards.
- Monitor the state disclosure patchwork. Requirements are emerging rapidly and vary in scope, and a single national or global campaign may trigger obligations in multiple jurisdictions.
- Prepare for EU AI Act transparency requirements. The August 2026 deadline for content marking and labeling obligations requires systems-level preparation, not just policy changes.
- Negotiate IP indemnities and require human sign-off on agentic tools. Assess the scope of indemnification obligations by ad platforms and AI vendors, including whether they indemnify for AI-generated content, and require human review before autonomous creative changes go live.
- Audit AI-generated imagery for accuracy and pair automated moderation with human review. Confirm that AI-generated or AI-modified product imagery accurately reflects the actual product or service (including any links) and build diverse human review checkpoints into moderation workflows.