In the first half of 2026, the crypto industry experienced 182 security incidents resulting in approximately $956 million in losses, but only about 12% of stolen funds were recovered. More alarming than the total losses is where the money went: according to SlowMist's analysis, funds were recovered or frozen in just 18 incidents, totaling approximately $118 million. The remaining nearly 90% of stolen funds are considered irretrievable.

The headline figures mask a troubling shift in how attackers operate. While losses fell nearly 60% year-over-year compared to H1 2025, this decline is almost entirely due to the absence of a singular mega-event like the previous year's Bybit incident, which cost approximately $1.5 billion. The actual number of incidents increased by about 50% year-over-year, revealing that attacks have not weakened; instead, they have fundamentally changed their target from exploiting smart contract code to exploiting human trust.

Why Are the Biggest Crypto Losses No Longer From Code Exploits?

The two most costly attacks in H1 2026 were not executed by finding vulnerabilities in smart contracts. Drift Protocol was drained of approximately $285 million through a six-month-long social engineering campaign that began with multi-signature approvals for transactions that "seemed harmless." In a separate incident, a victim in Singapore was invited to a video conference where government officials were entirely AI-generated, resulting in a loss of approximately SGD 4.9 million. These cases illustrate a critical insight: the most expensive vulnerability is now found in humans, not code.

When losses are analyzed by cause rather than incident count, the picture becomes even clearer. Supply chain attacks topped the list at approximately $298 million in losses, surpassing contract vulnerabilities at approximately $152 million and private key leaks at approximately $130 million. Cross-chain bridges, which facilitate asset transfers between different blockchain networks, accounted for only 20 incidents but caused approximately $346 million in losses, exhibiting what security experts call a pattern of "decentralized incidents, concentrated losses".

How Are Attackers Exploiting Supply Chains and AI Systems?

Supply chain poisoning has evolved across three dimensions: breadth, depth, and height. In one notable case, the Shai-Hulud worm published 637 malicious versions within 22 minutes, systematically stealing development pipeline permissions. In another incident, the security tool Trivy was compromised and became part of the attack chain itself, turning a defensive tool into an offensive weapon.

Attacks targeting AI Agents have shown a four-level evolution from the input layer to the supply chain layer. A typical case is the Bankr incident, where indirect prompt injection caused an AI Agent to recognize Morse-code-encoded instructions as legitimate transactions, resulting in approximately $175,000 in losses. This represents a new frontier where AI systems designed to automate crypto transactions become targets for manipulation.

Steps to Understand How Modern Crypto Attacks Work

• Social Engineering Infiltration: Attackers spend months building trust with employees or key stakeholders, gradually requesting seemingly harmless approvals that eventually drain protocol treasuries or user funds.
• Supply Chain Poisoning: Malicious code is injected into development tools, libraries, or security software used by crypto teams, allowing attackers to compromise projects at their source rather than targeting deployed contracts.
• AI Agent Manipulation: Attackers craft indirect prompts or encoded instructions that trick AI-powered trading or transaction systems into executing unauthorized transfers or approvals.
• Cross-Chain Bridge Exploitation: While bridges represent only a small fraction of incidents, they concentrate losses because they handle large volumes of assets moving between different blockchain networks.
• Money Laundering Infrastructure: Stolen funds are funneled through privacy protocols, cross-chain bridges, mixers, and theft-as-a-service tools to obscure their origin and prevent recovery.

Money laundering has become highly industrialized in the crypto space. Attackers like Lazarus leverage privacy protocols, cross-chain bridges, mixers, and theft-as-a-service tools to form a complete chain for moving stolen funds. This infrastructure makes recovery extraordinarily difficult, which explains why only 18 incidents in H1 2026 saw any funds recovered at all.

Most projects suffering the largest losses in H1 2026 had undergone professional security audits. The point of failure, however, lay precisely outside the scope of traditional audits: signing processes, cloud key management, validator nodes, and developer devices. An audit can prove a contract's logic is sound, but it cannot protect against a developer's compromised computer or a multi-signature holder being socially engineered.

What Are Experts Doing to Prevent These Attacks?

Some platforms and security firms have begun shifting their defensive strategies. OKX has protected approximately $526 million cumulatively by analyzing over 50,000 on-chain methods and managing signature risks. SlowMist has moved defense to the pre-action stage through built-in security execution gates and anti-money laundering screenings, attempting to catch suspicious transactions before they complete.

The fundamental challenge is that "trust" itself has become a new attack surface. In an era where AI-generated realities are possible, what needs to be verified is no longer a single piece of information, but the environment itself. This represents a paradigm shift in how the crypto industry must approach security, moving beyond code audits to include operational security, employee vetting, and transaction monitoring.

The data from H1 2026 paints a sobering picture: attacks are becoming more frequent, more sophisticated, and harder to recover from. The industry's focus on smart contract security, while still important, has become insufficient. As attackers increasingly target people and infrastructure rather than code, the crypto industry must fundamentally rethink how it protects assets and builds trust in an environment where verification itself has become uncertain.