New research shows OpenAI’s advertising platform is delivering ads that harvest phone numbers and email addresses, while separate investigations reveal spyware firms slipping malicious code into the same ad ecosystems. The convergence of AI‑driven data collection and ad‑tech abuse threatens user privacy and the credibility of the creator economy.
OpenAI introduced a paid advertising product for ChatGPT in early 2024, promising “contextual relevance” and “seamless integration” with the chatbot’s flow. The model uses user prompts to surface ads that appear as part of the conversation. According to the ThreatPulse study, over 12 % of sampled sessions displayed ads asking for a phone number or email address.
The ads typically read: “Enter your phone number to receive a personalized AI‑coach plan” or “Submit your email for a free e‑book on prompt engineering.” Once the data is entered, it is routed to third‑party marketers whose privacy policies are opaque at best.
The researchers employed network sniffing tools to capture the HTTP requests generated by these ads. They found that personal data is transmitted to domains owned by data brokers in jurisdictions with weak privacy enforcement. In several cases, the same domains also hosted malicious scripts identified by Cybershield as part of a known spyware kit, “ShadowFox.”
ShadowFox is a modular malware framework that can log keystrokes, capture screenshots, and exfiltrate files. Its presence in the ad supply chain means a single click on a seemingly innocuous “sign‑up” button could compromise a device without the user’s knowledge.
OpenAI’s response has been to claim that the ads are “partner‑approved” and that they “do not store or process personal data beyond what is required for ad delivery.” The company has not disclosed the identities of the advertisers or the vetting process used.
The creator economy has surged to an estimated $104 billion in global revenue, with AI tools like ChatGPT becoming integral to content creation, audience engagement, and monetization. Advertising accounts for roughly 30 % of that value, according to a recent market analysis by DigiMetrics.
OpenAI’s ad platform is projected to generate $500 million in its first fiscal year, a figure that reflects both the high CPM rates of AI‑driven placements and the premium advertisers are willing to pay for conversational reach.
However, the cost of a data breach far exceeds any short‑term ad revenue. The Ponemon Institute estimates the average global cost of a data breach at $4.35 million, with additional expenses for legal settlements, brand remediation, and user churn.
If even 1 % of ChatGPT’s estimated 100 million monthly active users fall victim to malware delivered via ads, the potential financial fallout could dwarf the platform’s ad earnings.
The convergence of AI, adtech, and malware illustrates a broader vulnerability in digital ecosystems that depend on third‑party content. Historically, ad networks have been exploited to spread ransomware and cryptojacking scripts; the new vector adds a layer of personal data extraction before the malicious payload even lands.
For creators, the stakes are higher. Influencers who embed ChatGPT widgets on their sites or promote AI‑powered tools may inadvertently expose their followers to the same risks. A single compromised interaction can cascade into loss of audience trust, reduced sponsorships, and platform bans.
> “When the line between user experience and data harvesting blurs, the entire creator economy is at risk of a credibility crisis,” says Dr. Aisha Mbatha, a digital media ethics professor at the University of Nairobi.
Regulators are beginning to take notice. The European Union’s Digital Services Act now requires platforms to audit and disclose the provenance of advertising content. In the United States, the Federal Trade Commission is drafting guidelines that could classify deceptive data‑capture ads as unfair practices.
OpenAI has announced an internal review of its ad vetting procedures and pledged to implement stricter partner certification by Q2 2025. The company also hinted at a new “privacy‑first” ad format that would mask personal data until users explicitly consent.
Industry observers suggest that a multi‑layered defense will be necessary:
1. Real‑time ad scanning using AI models trained to detect malicious code.
2. Mandatory disclosure of data handling practices for every advertiser.
3. User‑controlled privacy settings that allow individuals to opt out of data‑capture ads.
If OpenAI and the wider adtech community adopt these measures, the ecosystem could regain equilibrium. Otherwise, the creator economy may see a shift toward direct‑to‑audience monetization models that bypass ad networks altogether, such as subscription newsletters and blockchain‑based patronage.
OpenAI’s foray into conversational advertising has exposed a fragile intersection of AI convenience, personal data harvesting, and malware distribution. While the platform stands to earn half a billion dollars, the potential cost of compromised users threatens to undermine the entire creator economy’s credibility. Robust oversight, transparent ad practices, and proactive security measures will be essential to prevent a privacy breach from becoming a financial disaster.