Technique name: PromptPasta (Alethea-coined) Persona × scenario tags:

• CCO × competitor or adversary seeding misleading narratives into AI model outputs

• CCO × decontextualized information resurfacing in coordinated reply patterns

• CISO × coordinated inauthentic behavior using AI-generated content

• CISO × adversarial AI-output narrative manipulation

• Industry verticals: pharma, healthcare, finance, technology, regulated industries broadly

A new disinformation technique

While much attention has focused on positive applications of generative AI, adversaries are already exploiting the same tools for manipulation at scale. One of the most concerning techniques Alethea has identified is what Alethea calls PromptPasta.

PromptPasta is an AI-driven evolution of traditional copypasta tactics, where networks of inauthentic accounts use large language models (LLMs) to generate slightly varied but coordinated replies to high-visibility social media posts.

What Prompt

Pasta is

At its core, PromptPasta is the automated use of large language models to generate disinformation at scale. Traditional copypasta involves identical text copied and pasted by multiple accounts. PromptPasta produces subtly different variations of the same narrative using AI. This variation makes it much harder for automated detection systems to identify the coordination. To the average reader, the posts look like authentic, independent commentary. In reality, they are part of an orchestrated social media manipulation tactic.

How Prompt

Pasta works

PromptPasta campaigns typically involve:

The result is a flood of commentary that shapes online conversations before authentic voices can take hold.

Why Prompt

Pasta is dangerous

The danger lies in how effectively AI makes the posts look real:

• AI disinformation gains traction quickly, outpacing fact-checks or corporate responses

• Coordinated disinformation campaigns can portray companies as greedy, careless, or politically compromised

• AI manipulation — repeated false narratives can seep into training data, influencing future AI outputs

• Social media manipulation — coordinated replies tilt public perception, shaping what appears to be "authentic" grassroots sentiment

Where Prompt

Pasta has been used

In political discourse

Alethea's first report on PromptPasta documented a network of over 400 accounts amplifying narratives tied to the 2024 U.S. election and the second Trump administration. These accounts praised favored officials, attacked critics, and at times contradicted themselves — evidence of both the scale and the unpredictability of LLM-driven influence campaigns.

In pharma and healthcare

Alethea identified PromptPasta campaigns directly targeting pharmaceutical companies. At least 75 new accounts used AI-driven replies to push narratives around drug pricing, tariffs, and Medicare debates following the July 31, 2025 most-favored-nation pricing announcement. This marked a clear escalation: brands in highly regulated industries entered the crosshairs directly.

Alethea now assesses that PromptPasta LLM bot activity may exceed 1,000 active accounts, with at least 10,000 placeholders ready to be deployed following suspensions.

Why certain industries are especially at risk

Certain industries are especially vulnerable to AI disinformation and coordinated disinformation campaigns because:

• High regulation — policy changes, investigations, and compliance pressures can be triggered by public discourse

• Market sensitivity — stock prices and investor confidence swing on perception

• Public trust — reputation directly affects patient and consumer confidence and brand integrity

• Visibility — these companies are often at the center of heated political and social debates

In short: if people are talking about you, adversaries can weaponize that conversation.

Prompt

Pasta vs traditional copypasta

This evolution shows how generative AI lowers the barrier to running sophisticated manipulation campaigns.

Signs of Prompt

Pasta in action

Organizations should watch for:

• Clusters of accounts replying to high-visibility posts within minutes, with similar-but-not-identical messaging

• Reply-only behavior (no original posts) from accounts that appear active and legitimate

• Stolen or AI-generated profile imagery across the cluster

• Batch creation dates concentrated in tight time windows

• Occasional self-contradiction in the cluster's messaging — a tell of LLM-driven content production

Persona takeaways

For CCOs: PromptPasta is the named pattern when you see coordinated AI-generated reply activity attaching to an announcement, policy event, or news cycle that touches your brand. Detection at the network level — the batch-creation signatures, reply-only behavior, sub-minute response times — is what distinguishes a coordinated AI-driven amplification from organic debate.

For CISOs: PromptPasta is operationally relevant as a category of coordinated inauthentic behavior. The structural signatures (batch creation, reply-only behavior, sub-minute response times, contradictory messaging) are the kinds of indicators that detection systems and threat-intelligence workflows should be able to surface.

For investor-relations and regulated-industry teams: When a regulatory event is announced, expect PromptPasta-style amplification to attach within the same window. Monitoring narrative coordination alongside traditional sentiment analysis closes the gap.

Source links

• Insight summary — alethea.com/insights/promptpasta-ai-disinformation-narrative-attacks

• Original PromptPasta technique writeup — alethea.com/insights/promptpasta

• Pharmaceutical-industry application — alethea.com/insights/pharmai

• External coverage — alethea.com/insights/politico-promptpasta