From AI Darling to Disinformation Playbook: How Fake Accounts Managed a Market Frenzy and What DeepSeek’s Hype Teaches Leaders
DeepSeek, a new Chinese AI model, stormed app charts and briefly sent shockwaves through global markets. However, our disinformation research team found that much of this excitement was artificial — driven by thousands of fake profiles working in tandem. Their behavior bore the hallmarks of state-linked bot networks, amplifying hype and distorting market perception. For tech leaders and investors, the warning is clear: without the ability to separate genuine adoption from manufactured buzz, decisions risk being guided by disinformation rather than reality.
DeepSeek Disinformaton Campaign: Fake profiles used two main methods: amplifying each other to simulate popularity, and blending into authentic conversations to appear credible.
What happened with the DeepSeek launch?
TL;DR: DeepSeek’s record-breaking hype wasn’t organic — it was powered by thousands of coordinated fake accounts. #what-happened
Evidence:
Our disinformation research team analysed 41,864 profiles discussing DeepSeek.
3,388 were fake accounts — about 15% of all engagement , double the usual baseline.
These accounts generated 2,158 posts in a single day , reaching their peak activity.
The orchestrated amplification made DeepSeek trend across platforms and influence market narratives.
How did fake accounts manufacture excitement and drive disinformation around DeepSeek?
TL;DR: They acted in sync, boosting each other and hijacking authentic conversations to create an illusion of momentum. #how-manufactured
Fake profiles hijacked the DeepSeek hashtag to push scams and even promote competing AI platforms — exploiting hype for multiple agendas.
Evidence:
Mutual amplification: Fake profiles commented and liked each other’s posts to simulate popularity.
Piggybacking on trending posts: They flooded viral tweets (e.g., @FanTV_official, 480K+ views) with DeepSeek praise.
Integration with real users: Bots joined genuine conversations, tricking users into engaging with disinformation.
Synchronized timing: Bursts of identical content at the same moment maximised visibility.
What are the hallmarks of a coordinated bot network?
TL;DR: Recent creation dates, avatar recycling, identical posts, and synchronous activity are tell-tale signs. #criteria
Fake DeepSeek profiles exhibited classic bot behavior: identical, praise-filled posts, generic avatars, and synchronized timing.
Evidence:
Avatar recycling: Many used generic stock photos, often of Chinese women.
Recent creation: 44.7% of accounts were created in 2024 , coinciding with DeepSeek’s rise.
Copy-pasting: Identical praise-filled comments posted en masse.
Simultaneous posting: Coordinated bursts created artificial virality.
So what: These patterns match known behaviour of Chinese bot networks.
Why does manufactured hype matter for markets and brands?
TL;DR: Artificial excitement can sway investment, distort perception, and mislead decision-makers.
Evidence:
Market impact: DeepSeek’s hype briefly moved US markets, wiping out billions in valuation as investors scrambled.
Reputation distortion: Fake sentiment shaped the AI arms-race narrative in China’s favour.
Precedent: Similar tactics have been used to influence elections and protests — now redirected toward tech adoption.
Risk: Investors, boards, and communications teams risk making strategic decisions based on false signals.
Build vs. Buy: Can teams monitor this themselves?
TL;DR: Building internal detection is impractical; specialised tools are faster, broader, and more reliable. #build-vs-buy
Evidence:
Build: Requires data pipelines across platforms, AI expertise, and 24/7 monitoring.
Buy: Platforms like ours are pre-trained to identify fake accounts, coordinated behaviour, and narrative manipulation.
Time to value: Tools deliver insights in hours vs. months.
Cost of delay: One misinformation-fuelled hype cycle can move billions before teams react.
How should organisations roll out disinformation detection in 90 days?
TL;DR: Start with monitoring, then stress-test with simulations, and embed playbooks for rapid response. #rollout
Evidence:
0–30 days: Connect monitoring dashboards to social platforms; set thresholds for unusual spikes.
31–60 days: Run simulations of bot-driven hype or backlash; align comms and risk teams.
61–90 days: Develop and codify playbooks for investor messaging, market communications, and board reporting.
Roles: Comms & IR (messaging), Strategy (risk), Insights/Research (detection).
FAQ
Is all AI hype fake? No — but fake accounts can amplify genuine interest. How many fakes are “normal”? Typically 5–7%; DeepSeek hit 15%. Which platforms are most exploited? X and TikTok, due to virality and weaker controls. Are state actors always involved? Not always — but DeepSeek’s patterns matched Chinese bot networks. Can small firms be targeted too? Yes — hype tactics are cheap to deploy at any scale. Do detection tools replace analysts? No — they augment analysts with faster, deeper signals.
Methods & Data
Source: Our disinformation research team analyzed 41,864 profiles discussing DeepSeek between January 21 and February 4.
Findings: Identified 3,388 fakes (15%) showing hallmarks of coordinated bot networks.
Approach: Sentiment analysis, network mapping, and account authenticity scoring.
TODO: evidence needed — proposed investor survey on how hype narratives influence decision-making.
👉 Learn more about how our brand threat consulting and software services can help or enable you to identify coordinated fake accounts before they sway markets and reputations.