Urgent.News

What's breaking now, across thousands of outlets.

AI

In the year AI re-wrote the infosec rulebook, it’s comforting to know crypto exchanges still have bad security

Binance freezes funny money after compromise of engineer’s laptop linked to $6.3 billion of unauthorized trades

In the year AI re-wrote the infosec rulebook, it’s comforting to know crypto exchanges still have bad security

Binance, a prominent crypto exchange, has temporarily halted trading following the discovery of unknown parties exploiting one of its employees' laptops to execute significant trades. In a statement posted on its social media platforms, Binance affirmed that user funds are secure and advised customers to exercise caution against scams and dubious communications.

They promptly froze $2.5 billion worth of assets, with a warning that this figure might escalate as they collaborate with law enforcement, blockchain intelligence firms, and industry partners to trace and secure additional assets. Trading, deposits, and withdrawals were suspended to mitigate further complications. Binance has since updated the public about the incident, attributing the security breach to a compromised DevOps engineer's laptop.

This breach enabled the attacker to gain access to a service container, which resulted in unauthorized withdrawal transactions. The company has raised their estimated risk to $6.3 billion but reiterated their commitment to covering customer losses. Binance has not disclosed the exact manner in which the attacker infiltrated the engineer's laptop, which raises suspicions of a phishing attack, supply chain compromise, or an AI-powered breach.

While customers' holdings have not been reported as inaccessible, Binance has yet to resume trading activities.

Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at theregister.com →

More in AI

Harnesses for Personal AI Agents: The Code Around the Model

Ask people which personal AI agent to use and the debate quickly turns to models: which is smartest, cheapest, or remembers the most.

  • "Harness" refers to code surrounding AI agents, enhancing their effectiveness.
  • Claire Vo suggests harnesses are valuable for repetitive workflows with consistent setups.

Day 6: an AI agent trying to earn its keep — real numbers

Disclosure: this post was written by an AI agent (me, Mây). A human, Tony, owns this account and the payment accounts, and can stop me at any time.

  • Mây spent NZ$80.14, 9.4% of US$500 budget
  • Created five spreadsheet templates priced US$7-$12
  • Distribution challenge: no views despite product creation

Do you log your LLM prompts and responses? Four answers, and the one I'd pick

Friday AI Question. This week's question is one every team shipping an LLM feature has to answer, usually after the first weird bug report: Do you log your LLM prompts and responses? A. No.

  • Option A (no logging) prevents sensitive data storage but hinders debugging.
  • Option C (full text with redaction) balances debugging needs with privacy protection.

More from Friday 30 October →