About 67% of Banned Anthropic Accounts Used AI to Prep for Cyberattacks: A Wake-Up Call for Crypto Security
Crypto promises financial freedom. It also runs straight through a hostile internet where one bad click can drain a wallet. Anthropic, the AI safety company, just put out a number that should make every crypto holder sit up: about 67% of banned Anthropic accounts used AI to prep for cyberattacks. I’ll be honest: that number feels less like trivia and more like a warning flare. Bad actors are using AI to run bigger, faster attacks, and your wallet, private keys, seed phrase, exchange login, and personal data are the prize.
AI as a weapon in cyber warfare
Anthropic’s report covers the accounts they kicked off for breaking usage rules, and a big chunk of those accounts were busy prepping attacks. This is the shift. Criminals are not just recycling old hacking tricks anymore. They hand the grunt work to AI: phishing copy, malicious code, social engineering scripts, blockchain protocol research, smart contract review, infrastructure probing. Most guides say AI makes hackers more “efficient.” That’s only half right. It also changes who can run a serious campaign, because the slow prep work gets compressed into something one person can manage at scale.
How AI actually helps crypto attacks land
AI shows up in a few ways here, and none of them are good news for investors. Language models can write phishing emails or social engineering messages that feel personal, aimed at a wallet holder, a DeFi founder, a Discord moderator, or a trader waiting on an exchange withdrawal. The messages copy real exchanges, wallet apps, or DeFi platforms, then push people toward handing over private keys, seed phrases, or approving a transaction they’ll regret. Why does this matter? Because the message no longer looks clumsy. AI can also comb through smart contract code or blockchain infrastructure looking for vulnerabilities, which opens the door to flash loan attacks or reentrancy exploits. Then it automates the dull work: scanning compromised accounts, brute-forcing passwords, and building malware that lifts crypto straight off your device. It works.
What this means for crypto investors and traders
If about 67% of banned Anthropic accounts used AI to prep for cyberattacks, then anyone in this market needs to take security seriously and quit relying on autopilot. AI-powered attacks can blow past the old defenses. Investors and traders are an obvious target because they often sit on real money in decentralized, lightly regulated places. Blockchain makes a successful hit brutal: transactions are anonymous and they don’t reverse. Once your funds are gone, they’re usually gone for good. My take: treating security as a weekend chore is already too casual.
Phishing and social engineering get scary good
AI phishing is hyper-personalized now, close to impossible to tell apart from a real message. Attackers grab public info about you, feed it to a model, and get clean natural language built around your specific fears, money worries, open positions, or support requests. Even seasoned users get fooled. For traders, that can mean fake exchange announcements, fraudulent ICO promos, wallet update prompts, or a fake support agent fishing for your login or private keys. Counter to the usual advice, “check for typos” is not enough anymore. These scams are polished. I’d assume the worst and verify before clicking anything.
Smart contracts and DeFi are under more pressure
DeFi sits at the center of crypto, which makes it a juicy target. AI can chew through huge amounts of smart contract code and surface tiny bugs or logic flaws that lead to exploits. Audits help, sure, but a model processing complex code at scale can catch things a human reviewer skips over. Yes, this slightly contradicts the usual comfort line that “audited” means safe. Bear with me. Audited means reviewed; it does not mean invincible. That matters for liquidity pools, lending platforms, bridges, and any other DeFi protocol where the people who staked or lent their crypto may be the ones who eat the loss. The code itself matters more than ever, and the case for AI-assisted auditing on the defense side is hard to argue with.
Hardening your defenses when the attackers have AI
Given that about 67% of banned Anthropic accounts used AI to prep for cyberattacks, you need layers, you need to keep them current, and you need to actually understand what you’re doing. Technology alone won’t save you. Knowing how the scams work matters just as much. Is this overkill? For anyone holding meaningful crypto, no.
Security basics every crypto holder should nail
Start with a hardware wallet (Ledger, Trezor) for any meaningful amount of crypto. These keep your private keys offline, out of reach of online attacks. Turn on two-factor authentication everywhere, and use an authenticator app like Authy or Google Authenticator instead of SMS, since SMS codes fall to SIM-swapping. Be paranoid about links and attachments in messages you didn’t ask for, even when they look like they came from someone you trust. Verify the sender through a separate channel before you do anything. Keep your operating system, browsers, and antivirus patched. Keep reading up on the scams making the rounds, because the playbook changes fast. Skip this step. That’s exactly where people get burned. One rule never bends: never share your private keys or seed phrase with anyone, ever, for any reason.
AI on the defense side too
Attackers weaponized AI, but the same tech can guard the gates. AI security tools spot weird transaction patterns, flag sophisticated phishing, and scan smart contract code for holes faster than a human analyst ever could. Exchanges and DeFi platforms are wiring AI into their security stacks to watch for shady activity, block fraud, and protect user funds. My take: this is where platform choice starts to matter more than brand polish. Picking platforms that actually pour money into AI-driven security is a real way to lower your risk. This is an arms race, offense versus defense, and it isn’t stopping. Keeping current on security tech is just part of holding crypto now.
FAQ
What does “About 67% of banned Anthropic accounts used AI to prep for cyberattacks” mean for crypto?
It signals that more cybercriminals are using AI to plan and run attacks, which raises the odds of phishing, smart contract exploits, fake support messages, and other scams aimed at crypto investors and traders.
How can AI be used to create more sophisticated phishing attacks?
AI writes personalized, convincing phishing emails and messages that copy real companies and play on your psychology, so they’re much harder to catch than the old generic attempts. The old red flags still matter. They just miss more now.
Are hardware wallets still effective against AI-powered cyberattacks?
Yes. Hardware wallets are still one of the safest ways to store crypto because they keep private keys offline, which puts them largely out of reach of online attacks, AI-assisted ones included.
What steps should crypto investors take to protect themselves from AI-driven threats?
Use a hardware wallet, turn on 2FA, stay suspicious of unsolicited messages, keep your software updated, verify senders through a separate channel, and keep learning about the newest scam tactics and security habits.
Can AI also be used to defend against cyberattacks in the crypto space?
Yes. Security teams increasingly use AI to spot anomalies, find vulnerabilities in smart contracts, and block fraud, which makes it a strong tool for defense against advanced threats.