Advertisers Reframe Safety Around AI-Powered Ecosystem

"Brand safety" became what Business Insider calls a "Voldemort-like term" inside the ad industry after the Global Alliance for Responsible Media (GARM), which set brand-safety standards, dissolved in 2024 following an antitrust lawsuit from Elon Musk's X; a Texas federal judge dismissed that suit in March 2026, but the chilling effect persists, according to Business Insider. Embarrassing internal GARM emails surfaced through a Republican-led congressional investigation and X's litigation, and industry insiders now doubt a successor group will emerge. Jamie Barnard, Unilever's former general counsel for media and now head of privacy-tech firm Compliant, told Business Insider that oversight of AI in advertising will likely fall to buyers rather than regulators: "So who's going to regulate? It's the buyer." A media agency executive involved in cross-industry AI-safety talks said messaging is being kept deliberately "methodical," avoiding flashpoints such as bias in AI-powered ad algorithms, aiming for something "as antiseptic and apolitical as possible."

Three standards bodies moved to fill the gap this year: the Media Rating Council released an ad-auction transparency framework for machine-learning-driven bidding, the Interactive Advertising Bureau published a March white paper recommending guardrails for agentic AI in video ads, and the International Chamber of Commerce issued its own guidance on responsible AI use in marketing, per Business Insider. Meanwhile, political ad spending is colliding with AI directly: PAC-backed groups are running ad campaigns to shape AI regulation ahead of the midterms, according to The Hill, and AdImpact projects total 2026 midterm political ad spending will hit a record $10.8 billion (later revised toward $11.6 billion), roughly 20% above 2022. In one high-profile example, the National Republican Senatorial Committee released an AI deepfake ad of Texas Senate candidate James Talarico reading fabricated statements, which CNN reported a UC Berkeley forensics expert called "hyper-realistic" enough to deceive most viewers despite an on-screen "AI GENERATED" disclosure.

Separately from the politics, marketers face a harder detection problem: legacy brand-safety tooling was built to filter known categories like adult content and hate speech using blocklists and contextual classifiers, but generative AI produces convincing "slop" sites and inserts ads inside conversational chatbot output in ways that can defeat those older signals, according to Campaign Live and Digiday. Google and other platforms are deploying their own AI to fight invalid traffic, per eMarketer, while Senator Ed Markey has separately pressed AI chatbot makers including OpenAI over the privacy and safety implications of in-chat advertising, according to Forbes, and Anthropic has run ad campaigns positioning itself against rivals it suggests will add chatbot ads, per The Guardian.

Two separate workstreams are converging: legal and communications teams need to keep AI-safety messaging deliberately narrow and apolitical to avoid GARM's fate, while ad-ops and data teams need genuinely new detection signals - generative-content classifiers, conversational-ad provenance tracking, and auction-transparency reporting aligned to the new MRC/IAB/ICC frameworks - since pre-AI blocklists and classifiers won't catch AI-native ad fraud or placements.

Whether any group attempts to reconstitute GARM's brand-safety standard-setting role, how the IAB's and MRC's new frameworks get adopted by DSPs and SSPs, whether more AI deepfake ads appear in competitive midterm races beyond Texas, and whether the FTC or Congress moves on AI-chatbot advertising disclosure following Markey's inquiry.