Skip to content
Est. MMXXVIVol. VI · № 273RSS
Blockchain Breaches

An archive of cryptocurrency security incidents — hacks, exploits, bridge failures and rug pulls, documented with on-chain evidence.

Dossier № 209Phishing / Social Engineering

Banana Gun Telegram Oracle Drain

Telegram message oracle flaw let an attacker drain $3M from 11 Banana Gun users via manual transfers on victim wallets. Team refunded victims from treasury.

Date
Chain(s)
Status
Recovered

On September 19, 2024, the popular Telegram-based crypto trading bot Banana Gun suffered a highly targeted exploit that drained approximately $3 million from 11 user wallets — most of them experienced "smart money" traders. The attackers exploited a vulnerability in the Telegram message oracle that the bot used to validate user commands. Banana Gun fully refunded affected users from corporate treasury.

What happened

Banana Gun was, by September 2024, one of the larger Telegram-based DeFi trading bots, having processed over $6.3 billion in trading volume across nearly 279,000 users. The bot operated by:

  1. Receiving user commands through Telegram chat (buy this token, sell that one, etc.).
  2. Verifying the command via Telegram's API.
  3. Executing the corresponding on-chain transaction from a user's bot-managed wallet.

The attack exploited the second step. The Telegram message oracle — the bot's mechanism for proving "this command came from the legitimate Telegram user" — had a vulnerability that let the attacker inject crafted messages that the bot interpreted as legitimate transfer commands from the victim, even though the victim had not sent them.

The targeting was distinctive:

  • 11 victims, mostly experienced traders — not random users. The attacker had clearly profiled the bot's user base and selected high-value targets.
  • Manual transfers rather than automated drains — each attack required specific message construction per victim, suggesting hands-on operation rather than a generic exploit script.

Total extracted: ~$3M, primarily in ETH and major tokens held in the victims' Banana Gun-managed wallets.

Aftermath

  • Banana Gun paused both EVM and Solana bot operations within hours.
  • The team publicly committed to full refunds for all affected users from treasury reserves, explicitly noting that no BANANA tokens would be sold to fund the reimbursement (which would have hit token holders).
  • Operations resumed with an added 6-hour transfer delay on outgoing transactions — a circuit breaker designed to give the bot or its users time to detect anomalous behaviour before funds leave the wallet.
  • BANANA token price rose 7% on the refund announcement, reflecting market confidence in the team's response.

Why it matters

The Banana Gun incident is a case study for the emerging risk class of Telegram trading bots — products that have grown to multi-billion-dollar trading volume by 2024-2026 but whose security models depend on:

  • Telegram's authentication and API integrity (the bot can't really verify the user beyond what Telegram tells it).
  • The bot operator's wallet-management infrastructure (often hot-wallet-style key custody for fast trading).
  • The bot's own code quality (often written and updated rapidly to keep pace with memecoin-trading demand).

Any of those three layers being compromised exposes user funds. Banana Gun's specific vulnerability was in the oracle layer; similar bots have suffered different-layer issues at smaller scales.

The structural lessons:

  1. Trading-bot custody is a meaningful asset class in the threat model — distinct from self-custody, distinct from exchange custody, with its own unique vulnerabilities.
  2. Targeted "smart money" attacks are increasingly worthwhile when attackers can identify high-value users — the same Inferno-Drainer-style profiling that hit the Whale Hunter's Payday victim applies here.
  3. 6-hour transfer delays are a meaningful UX/security trade-off — they reduce attack-window-to-exploit time but slow legitimate user operations. Most users will accept the trade-off after an incident; the question is whether the bot ships with the delay before or after the first incident.

Banana Gun's response — fast pause, full refund, immediate UX hardening — set a credible bar for how a Telegram-bot operator should handle a security incident at this scale.

Sources & on-chain evidence

  1. [01]cryptonews.comhttps://cryptonews.com/news/telegram-bot-banana-gun-to-refund-3-million-hack-victims/
  2. [02]quillaudits.comhttps://www.quillaudits.com/blog/hack-analysis/banana-gun-exploit
  3. [03]bravenewcoin.comhttps://bravenewcoin.com/insights/telegram-bot-banana-commits-to-covering-us3-million-lost-in-hack

Related filings