Binance Co-CEO Richard Teng Says Security Remains Core to Crypto Trust
Binance wallet intercepted $540 million in potential losses in the first half of the year.
Binance Co-CEO Richard Teng stated that security remains the core of crypto trust. In the first half of 2026, the Binance wallet security center helped users avoid approximately $540 million in potential losses and reminded that AI makes attacks more sophisticated; spending a few extra seconds to verify can change the outcome.
The center, launched on February 3, conducts real-time scans for keyless multi-party computation wallets and imported mnemonic wallets, covering link openings, transaction signatures, and unknown token deposits. Official data also includes filtering approximately 206 million spam transfers, covering about 2.6 million users, identifying around 4.93 million high-risk transactions across 19 chains, and discovering about 996,000 instances of malicious authorization.
Changes on the attack side show that malicious entities have nearly tripled quarter-over-quarter. AI is being used to generate malicious code, phishing sites, and fake identities in bulk, shifting from broad net tactics to targeted fraud, including "buy only, no sell" and hidden backdoor contracts. The security side also uses models to analyze token behavior, site risks, and attack intentions, providing one-click revocation of authorization. SlowMist conducted an audit of its design in February.
In industry comparisons, CertiK's Hack3D report for the first half of the year indicated that phishing incidents decreased by over 50% year-on-year, but losses only dropped by about 10.8%, still around $370 million. Binance's open letter to users at the beginning of the year stated that under the proof of reserves, user assets were approximately $162.8 billion. Red team exercises reduced the internal phishing success rate from 3.2% to 0.4%, and assisted about 50,000 external scam victims in recovering approximately $11.7 million.
In market mechanisms, buyers are retail wallet users afraid of having their signatures drained, while sellers are exchange wallets that make security a product entry point. The driving force behind incidents comes from the expansion of fraud capacity after AI reduces costs, with funds remaining in user addresses from intercepted authorizations and phishing transactions, rather than flowing to attack contracts. Benefiting are platforms that can make risk control a default scan, while under pressure are attack groups relying on bulk fake sites for customer acquisition, and users who still treat "multiple clicks for authorization" as a habit.
Teng also tied the security narrative to 24-hour trading and perpetual contract expansion: the market never stops, and attacks do not take a break, making wallet interception a supporting capability for exchanges competing for custody of external assets.
Source: Public Information
ABAB AI Insight
Since Richard Teng took over Binance, his public actions have consistently framed compliance and security as tickets for regaining customers, rather than technical appendices. The wallet security center has turned interception figures into a communicable trust asset, similar to how exchanges use proof of reserves to address concerns over bank runs: first providing a headline-worthy dollar figure, then embedding the product into every signature.
Capital movement occurs at both ends. On one end, budgets for audits, on-chain scanning, and customer service recovery are front-loaded to wallets, reducing users' fear of default when keeping assets on-chain; on the other end, the same set of interception data is used to counter the narrative that "centralized platforms are unsafe," bringing Web3 traffic back to Binance's account system. The motivation is that after the growth of external trading, the platform must prove it can still survive when users leave the order book.
Analogous actions include Apple removing fake Ledger applications, Circle promoting hardware multi-party keys, and various companies making "revoke authorization" an independent entry point. The industry is in a phase of productizing defenses after the industrialization of attacks: while phishing incidents have decreased, the stickiness of single-transaction losses remains, indicating that higher net worth and more targeted strikes are still slipping through the cracks.
Structural judgment belongs to technological substitution: AI simultaneously arms both offense and defense, with victory no longer determined by contract audit reports, but by the default interception in the first three seconds before signing. The mechanism is a collapse of marginal costs—after the costs of fake sites and deep fakes approach zero, broad-spectrum attacks become ineffective, and platforms must shift risk control from post-event compensation to preemptive blocking, or else retail trust will collapse before token prices do.
ABAB News · Cognitive Laws
Trust is not a slogan; it is the dollars intercepted.
After the cost of attacks approaches zero, defense must become the default key.
Those who wait three more seconds deserve to hold private keys.