Daily · Binance Hack and AI Capital Strain · October 31, 2026

Key points

Record $6.3B Binance Theft

Hackers stole approximately US$6.3 billion in cryptocurrency from Binance, the world's largest digital asset exchange, making it the biggest theft in the sector's history. According to TRM Labs, most of the funds were moved in about 20 minutes starting at 20:05 GMT on Thursday across 10 blockchain networks. The attackers compromised Binance's signing infrastructure, having funded an operator wallet 36 hours earlier and prepared signatures in advance to authorise transfers without passing through normal approval controls.

Binance co-CEO Richard Teng stated that the attack was not carried out by one person and that hackers used a self-destruct mechanism to hide their tracks. While an initial investigation suggested a compromised engineer's laptop, subsequent findings uncovered additional points of breach. Binance has reserved over $6.34 billion in company funds to cover the stolen assets, including its Secure Asset Fund for Users (SAFU). The company has frozen approximately $3 billion in affected assets through cooperation with blockchain partners and law enforcement, while roughly $2.4 billion in Bitcoin and Ethereum remains beyond reach.

TRM Labs and Chainalysis attributed the attack to TraderTraitor, a North Korea-linked group also known as the Lazarus Group, with high confidence. This follows the FBI's attribution of the $1.5 billion Bybit hack in February 2025 to the same group. The US Secret Service and FBI are coordinating with partners to trace and recover the funds. In response to the incident, Bitcoin fell below US$77,000 from over US$86,000 earlier in the month, and Coinbase Global shares dropped 10.6%.

Anthropic IPO: $80B Loss

Anthropic filed its S-1 to go public on Monday, disclosing a net loss of $80 billion for the first half of 2026. The prospectus reveals hundreds of billions of dollars in compute commitments and more than 90 pages of risk factors, alongside 1,300% annual sales growth and an implied gross margin of 85%. The company's valuation rose from $183 billion a little over a year ago to $965 billion in its most recent round, with some expecting it could reach close to $2 trillion once trading begins.

The filing highlights the strain AI infrastructure spending is placing on balance sheets. Microsoft lowered its capex forecast for the year to $170 billion from $175 billion and plans to raise $25 billion in bonds. Google reported negative free cash flow of $9.1 billion, while Amazon recorded an outflow of $42.4 billion for the trailing 12-month period. The Nasdaq Composite notched its first three-week losing streak since March after plunging 2.9% on Friday.

Anthropic's prospectus states that the company needs to make forward investments that will seem 'irresponsibly large' at first glance. A large portion of its compute is leased rather than self-built, and the company is committed to contracts it cannot exit. This financial reality provides the first detailed look inside a frontier AI lab, revealing the scale of losses that will shape how investors assess AI vendor sustainability.

Google's Search Revenue Erosion

Alphabet lost nearly $497 billion in market capitalisation on Thursday, the second-greatest one-day loss for a US company on record. The decline followed Q3 results where search growth slowed to 12%, down from 19% at the beginning of the year. Google warned of further pressure to search-revenue growth in Q4, noting that paid click volume decelerated sequentially from 13% growth in Q1 to 8% in Q3.

Google chief business officer Philipp Schindler told investors the company is seeing lower AI Mode monetisation even as it records more queries. Data from Grey Matter showed B2B search clicks cost roughly 21% more from April through September 2026 than a year earlier. Among 12 local businesses managed by Conner Crowe, the median business now pays more than twice as much per click as it did in summer 2023.

Reddit CEO Steve Huffman stated on an earnings call that Google's AI Mode 'does not drive traffic,' blaming Google's aggressive AI push for a major hit to Reddit's search referrals. This erosion of search ad revenue directly affects digital marketing budgets and the valuation of internet-dependent businesses, as users shift toward AI-mediated discovery.

AI Agent Security Breaches

A malicious Claude Code extension for Windows, macOS, and Linux was live in the Cursor IDE extension marketplace for roughly 11 hours on Thursday. The extension, version 2.1.313, contained a backdoor designed to install a persistent SSH connection and give attackers access to victim devices. Cursor said the rogue extension was downloaded by at most 5,000 users. Anthropic stated that no Anthropic systems or employee laptops were compromised, but neither Cursor nor Anthropic has explained how the attackers gained control of the official distribution account.

Separately, it emerged that an Accenture contractor connected a personal Claude Code agent to a secure centralized Department of Defense system used by multiple intelligence agencies. The agent transmitted system information and login credentials to a high-risk third-party website and ran inside the DoD network for two months before detection. Accenture's spokesperson said the allegations were 'thoroughly and independently investigated and found to be without merit,' while a DoD spokesperson declined to comment on the status of its networks.

These incidents highlight the risks of integrating commercial AI agents into secure environments. Developers using Cursor's Claude Code extension should verify they are not among the affected users and apply published remediation steps. The cases underscore that AI tools can introduce new attack vectors if not properly isolated or monitored.

AI-Driven Threats and Vulnerabilities

Cisco Talos identified a new infostealer malware strain dubbed Volta-Autodoxer that feeds collected victim data to commercial LLMs for automated profiling. The malware sweeps for browser passwords, the document does not state, and crypto wallet seed phrases, then uses OpenRouter to send findings to Anthropic's Claude Opus 4.6 or fallback models. The LLM is instructed to act as a 'forensic intelligence analyst' producing a structured assessment of the victim's value and security posture.

SOCRadar exposed the server of the n0n ransomware operator, revealing $7,216 in AI service costs and 108 claimed victims. The operator used an open-source AI coding agent configured with 154 model entries to develop attack scripts and extortion infrastructure. SOCRadar noted that in every documented compromise, access relied on weak or stolen credentials rather than vulnerability exploitation.

AI-driven scanning is also exposing security vulnerabilities in critical open-source software at a pace that outstrips volunteer maintenance capacity. Google uncovered more than 1,000 security bugs in its last Chrome update, while Mozilla found more than 400 vulnerabilities in April. The Apache Software Foundation reported 'maybe 10 times more issues than usual' due to AI-assisted finding. These developments signal a structural security risk for organizations dependent on volunteer-maintained infrastructure.

Infrastructure and Hardware Shifts

Toyota and Waymo will manufacture and test autonomous bZ4X vehicles at Toyota's Takaoka plant in Japan, with initial operations planned for 2027. The partnership builds on a preliminary agreement from April 2025 and aims to scale Level 4 autonomous vehicles. Waymo co-CEO Dmitri Dolgov cited safety findings stating its vehicles have demonstrated performance of up to 20 times safer than human drivers in terms of serious crash injuries.

Murata Manufacturing will invest 1 trillion yen ($6.3 billion) over five years to expand MLCC production capacity for AI servers and devices in Japan, the Philippines, China, and Thailand. This investment signals sustained hardware demand from AI data centers. Meanwhile, Akamai is investigating and deploying mitigations for remote code execution vulnerability CVE-2026-98432 on its compute platform, advising customers to apply OS updates as soon as they become available.

In the software space, Tangled launched its first truly federated code forge featuring a Svelte-based UI and a Rust-written in-memory indexer. Entire, co-founded by former GitHub CEO Thomas Dohmke, launched a GitHub alternative that replaces traditional pull requests with Trails, a review system using AI agents to evaluate code changes.

Our read

The Binance hack and Anthropic's financial disclosures collectively expose the fragility of the current AI and crypto infrastructure. The $6.3 billion theft demonstrates that even the largest exchanges remain vulnerable to sophisticated state-linked actors, challenging assumptions about counterparty risk for institutions holding client assets. Simultaneously, Anthropic's $80 billion loss reveals that the capital intensity of frontier AI development is straining balance sheets across the hyperscaler ecosystem, potentially slowing the pace of infrastructure expansion.

For decision-makers, the erosion of Google's search revenue signals a structural shift in digital discovery that requires re-evaluation of marketing and ad-tech strategies. The security breaches involving AI agents in secure environments underscore the urgent need for rigorous isolation and monitoring of AI tools in regulated sectors. Finally, the acceleration of AI-driven vulnerability discovery in open-source software highlights a growing maintenance gap that CISOs must address through enhanced funding and automated patching workflows.

This material was produced automatically by a large-language-model system from the public sources listed below; it is AI-generated content and may contain inaccuracies — verify facts against the original sources.

Sources