Key Takeaways
In recent years, X, formerly Twitter, has become the main stage for crypto discourse and, increasingly, for crypto scams.
Botnets, impersonators, and spammy “arbitrage bot” schemes flood feeds with misleading links and fake token promos. Add AI-generated content and tactics designed to game the algorithm, and you have a hostile environment for ordinary users and legitimate builders alike.
This article explains how scams work, why they persist, what X says they are doing, and how users can protect themselves. It is augmented by expert commentary from security practitioner Daniel Brundson, Growth + Local Lead at human.tech by Holonym.
+76
Not all bad actors on X are fully automated. Many accounts are semi-automated or operate in coordinated clusters, blending human operators with scripted posting to amplify reach.
Impersonation bots pose as influencers or project founders to lend credibility to phishing links. Meanwhile, AI/LLMs help botnets produce more natural-sounding replies, masking repetition and obvious tells that automated detection tools used to catch.
Brundson told CCN: “X has been flooded with bots promoting crypto scams. From your perspective, why has the platform failed to control this problem?”
“X does have tools to address spam, but the individuals responsible for the spam are smart. They understand the X tools, and are able to quickly make adjustments to work around the tools, creating a game of whack-a-mole that has been lucrative enough for the spam creators that they continue to invest time, money, and effort into the strategy.”
AI lowers the cost of producing convincing content and increases the variability of posts across a botnet, making pattern-based takedowns harder.
Brundson states, “How are these bot-driven scams evolving: are they just spam, or are they now using AI to appear more human?”

“AI has definitely made it easier for bots to appear more human and work around X’s defenses. Spam-prevention tools that are looking for specific patterns to take down bot networks are going to have a more difficult time identifying and applying rules that can target large spam networks due to the variation that they can add to their posts using AI,” Brundson added.
Common patterns include fake giveaways, “airdrop” announcements that link to wallet drainers, and arbitrage bot contracts that claim to run profitable strategies but actually siphon funds once you approve permissions.
One tactic researchers and journalists have flagged is reply-and-block: bot accounts reply to your post to register engagement and then immediately block you, possibly signaling to ranking systems that your content should be de-boosted.
For Brundson, who spent several years at Twitter (now X), “Some reports suggest bots reply-and-block to “deboost” real accounts. How dangerous is this kind of algorithm manipulation for public trust and what role human.tech plays to protect users from this kind of scam?”
“I’m unfortunately not familiar with this strategy. However, if X were to enable users to link their blockchain wallets to their accounts, we’d be able to apply a human verification mechanism, similar to the previous verification system that Twitter used for both reputable accounts and bots, but at scale.”
Brundson also said that any malicious strategy to reduce the visibility of reputable human accounts reduces the quality of the platform, as key voices in important areas might have a difficult time communicating perspectives that add value to, or are critical to shape, the public conversation.
The broader harm goes beyond individual creators — it undermines the integrity of discourse itself: throttling authentic voices and amplifying fake ones erodes trust.
X has rolled out measures to deter spam, including a “Not A Bot” program that charges a small fee for new accounts to interact, and periodic sweeps that suspend networks violating manipulation rules. The company also touts legal action against groups attempting to bribe their way back onto the platform.
But critics argue these steps don’t materially raise the cost for determined operators; fees are trivial at scale, and enforcement can be inconsistent and opaque. Meanwhile, reduced data/API access has made independent measurement harder.
Human.tech’s Brundson emphasizes that “Do you think Elon Musk’s paywall experiments and the “Not A Bot” program are real solutions, or just PR?”
“Elon enabling anyone to verify their account with a paid subscription removed any value the verification program previously had. Making it pay-to-play may have increased their revenue, but made it much more difficult to understand who is behind accounts, especially high-profile figures.”
For the expert, they have made progress in reinstating a version of verification that does help to identify different types of accounts, but bad actors can still appear to be credible with a simple subscription.”
The immediate harm is financial—people lose funds to wallet drainers, fake token promos, or “arbitrage bot” contracts. But the second-order effect is reputational: when scams trend louder than innovation, the entire field suffers.
Brundson said: “Bad news will always find good publicity coverage. The more scammers see success, the more they will be featured in the news and further drive a negative perception of the space, while positive developments are buried.”

According to him, this damages the entire industry, preventing new users from onboarding, and money from coming in to help develop positive use cases.
For projects trying to build, bot-amplified noise crowds out genuine updates. For the public, nonstop exposure to scams fosters the belief that Web3 is a grift, even as credible teams continue to ship.
Experts point to two broad fixes: rebuild trust and safety capacity and raise the cost of inauthentic behavior, not just with fees but with better verification and more transparent enforcement.
About what would be the most urgent step to cut down bot-driven crypto fraud today, Brundson said: “Invest in your trust and safety team. They were well funded and doing great work before the acquisition, but the majority of the team was let go as part of the acquisition.”
“I’m not sure how well that team is funded now, but they can use funding to help research and implement new tooling to address the problem.”
“Also, consider new human verification methods. There are many out there, and they could help users better understand the context of the account they are engaging with,” he added.
Concrete avenues often cited by researchers and practitioners include:
Until systemic fixes land, personal security hygiene is your best defense:
X’s crypto conversation is robust and precarious. Botnets and scammers have seized algorithmic quirks and verification gaps to scale their operations, while AI helps them blend in. Platform moves like small paywalls or sporadic purges haven’t changed the economics of abuse enough to stem the tide.
Daniel Brundson’s perspective underscores the core truth: this is whack-a-mole because it’s profitable and defenses haven’t evolved as quickly as offenses.
Reinvesting in trust and safety, augmenting with human-centric verification, and restoring measurement transparency are the practical steps that can shift incentives.
Until then, user vigilance is the firewall. Stay skeptical of viral claims, slow down before you click or sign, and lean on verified sources. In a market where attention is currency, your caution is capital.
After Elon Musk’s acquisition, X loosened some moderation and verification controls while becoming the primary hub for crypto discussion. That combination of high visibility, low friction, and reduced safety oversight created fertile ground for scammers. Yes. AI and large language models make bots more adaptable and human-like. This tactic involves bots replying to a user’s post to register engagement, then immediately blocking them, potentially signaling the algorithm to suppress that user’s content (“deboosting”). It’s a manipulation strategy that harms visibility and trust on the platform. Financial losses are the most direct consequence; users lose funds to drainer contracts or fake tokens. But the broader damage is reputational.