Crypto hack counts fall but supply chain attacks reshape threat landscape

cointelegraphPublished on 2025-12-23Last updated on 2025-12-23

Abstract

New data from CertiK reveals that while crypto hackers stole $3.3 billion in 2025, the number of attacks fell sharply. Losses were concentrated in fewer, more damaging supply-chain attacks, which accounted for $1.45 billion across just two incidents, including the $1.4 billion Bybit hack. This shift indicates attackers are moving away from simple code vulnerabilities toward more sophisticated infrastructure-level exploits. The number of security incidents decreased by 162 year-over-year, suggesting improved protocol-level security. The median loss per hack fell 35.75% to $103,966, though the average loss rose to $5.3 million due to high-value outliers. Phishing scams were the second-largest threat, costing $722 million across 248 incidents. A significant subset was "pig butchering" romance scams, which used prolonged emotional manipulation and cost the industry $5.5 billion in 2024.

Crypto hackers stole $3.3 billion in 2025, but the number of attacks fell sharply as losses became concentrated in fewer, more sophisticated supply-chain exploits, according to new data from blockchain security firm CertiK shared with Cointelegraph.

While total losses remained elevated, the decline in incident counts and a drop in median theft sizes suggest that protocol-level security is improving, pushing attackers away from simple code vulnerabilities and toward phishing and infrastructure-level attacks.

CertiK said supply-chain breaches emerged as the most damaging threat, accounting for $1.45 billion in losses across just two incidents, including the $1.4 billion Bybit hack in February.

"The Bybit exploit signals that well-capitalized, well-coordinated threat actors are becoming more active across the ecosystem," the report said, predicting a rise in the “sophistication” of supply chain attacks as attackers target more infrastructure providers.

Crypto hacks by amount and incident, yearly chart. Source: CertiK

Related: Soulja Boy token sparks backlash after Base co-founder posts purchase receipt

The number of security incidents decreased by 162 counts year-over-year, indicating that blockchain cybersecurity measures are improving despite hackers aiming for larger targets.

The average amount lost per hack stood at $5.3 million, a 66% increase from the previous year. However, the median loss — a measure less influenced by outlier incidents — fell to $103,966, down 35.75% over the same period.

Cryptop hacks by incident type and amount of losses, one-year chart. Source: CertiK

Related: Solana AI token Ava hit by launch sniping tied to deployer: Bubblemaps

Code vulnerabilities fade as “pig butchering” scams threaten crypto savings

Phishing scams became the second-largest threat, costing crypto investors a cumulative $722 million across 248 incidents.

Recently, an investor lost their entire Bitcoin (BTC) retirement fund in an artificial intelligence-fueled romance scam, also known as a "pig butchering" scam, where the con artists used prolonged emotional manipulation to convince the investors to transfer their funds.

Pig butchering victim stats, grooming time. Source: Cyvers

Pig butchering scams are a subset of phishing scams that cost the industry a collective $5.5 billion in 2024, across 200,000 individual cases.

Notably, the average grooming period for victims is between one and two weeks in 35% of cases, while 10% of scams involve grooming periods of up to three months, according to blockchain security platform Cyvers.

In June, the US Department of Justice announced the seizure of over $225 million in crypto linked to pig butchering scams.

Magazine: Coinbase hack shows the law probably won’t protect you — Here’s why

Related Questions

QAccording to CertiK's data, what was the total amount stolen by crypto hackers in 2025 and what was the most damaging type of attack?

ACrypto hackers stole a total of $3.3 billion in 2025. The most damaging type of attack was supply-chain breaches, which accounted for $1.45 billion in losses.

QWhat does the report suggest about the trend in protocol-level security based on the decline in incident counts and median theft sizes?

AThe decline in incident counts and the drop in median theft sizes suggest that protocol-level security is improving. This is pushing attackers away from simple code vulnerabilities and toward more sophisticated methods like phishing and infrastructure-level attacks.

QWhat was the average amount lost per hack and how much did it change from the previous year?

AThe average amount lost per hack stood at $5.3 million, which was a 66% increase from the previous year.

QWhat are 'pig butchering' scams and how much did they cost the industry in 2024?

A'Pig butchering' scams are a subset of phishing scams that involve prolonged emotional manipulation to convince victims to transfer their funds. They cost the industry a collective $5.5 billion in 2024 across 200,000 individual cases.

QWhat significant action did the US Department of Justice take regarding pig butchering scams in June?

AIn June, the US Department of Justice announced the seizure of over $225 million in cryptocurrency that was linked to pig butchering scams.

Related Reads

How to Do Research Well: Deliberately Practice the Real Skills That Matter

No one truly teaches you how to do research. You're often given a desk, a pre-selected problem, and vague instructions to "create something new." Consequently, many people reverse-engineer the job based on visible outputs—papers, posts, announcements—learning only how to *appear* like a researcher rather than how to *become* one. True research capability is built from stacking small, trainable skills, nearly all of which can be developed through deliberate practice. **Pick Your Own Problem:** Most researchers absorb problems from advisors or trends, lacking the underlying reasoning. Choosing a problem you genuinely care about, as John Schulman advises, leads to original work. Develop "taste" like a muscle: predict experiment outcomes, guess paper results from methods, and track which findings remain important over time. **Upgrade Your Inputs:** Relying on shared reading lists (arXiv hot lists, filtered group chats) leads to unoriginal conclusions. Undervalued old literature often holds crucial insights (e.g., MoE, LSTM, backpropagation). Richard Sutton's "The Bitter Lesson" or Claude Shannon's 1952 talk on creative thinking are more predictive than lengthy modern surveys. Breadth matters as much as depth: draw from neuroscience, mechanism design, hardware knowledge, and honest statistics. Read papers directly, especially appendices and limitations sections. **Write Everything Down:** As Paul Graham noted, writing exposes flaws in seemingly mature ideas. Writing is the cheapest defense against self-deception. Following Feynman's principle, Darwin programmatically wrote down facts contradicting his theory to combat memory bias. Maintain a detailed log of hypotheses, setups, predictions, results, and updated understandings. Reviewing past logs fosters essential humility.

marsbit1h ago

How to Do Research Well: Deliberately Practice the Real Skills That Matter

marsbit1h ago

Following US Ban on Fable 5, Zhipu AI's Stock Soars 47%

On June 15th, shares of Zhipu AI surged dramatically on the Hong Kong stock market, peaking at a 47.6% gain before closing 32.82% higher. This sharp increase was directly triggered by two recent industry events. On June 12th, Anthropic announced it was suspending global access to its latest flagship models, Claude Fable 5 and Claude Mythos 5, to comply with a U.S. government export control order. The next day, Zhipu AI announced it would open access to its latest open-source flagship model, GLM-5.2, under the permissive MIT license. The Anthropic incident highlighted a critical issue beyond raw model capability: the risk of sudden, unpredictable loss of access to advanced AI models, especially for developers and enterprises deeply integrated with them. This has shifted industry and market focus toward factors like stability, sustainable access, and controllability. Zhipu's move, promoting "frontier intelligence for all," positions its openly available model as a reliable and accessible alternative. The GLM-5.2 model emphasizes "Long Horizon Task" capabilities with a 1M context window, targeting complex, multi-step coding and engineering workflows where maintaining context is crucial. Analysts note this event exposes the risk of dependency on closed-source models subject to single jurisdictional controls, potentially accelerating a shift toward domestic base models and localized deployments. The market's reaction signals a new valuation dimension in AI: providers who can offer stable, long-term, and sustainably accessible AI capabilities are gaining strategic importance.

marsbit2h ago

Following US Ban on Fable 5, Zhipu AI's Stock Soars 47%

marsbit2h ago

Fully Entering the AI Era: Alipay Bets on Conversation, WeChat Holds Fast to Social

In May 2026, Alipay announced over 300 million AI payment transactions. Shortly after, WeChat opened its mini-programs for AI integration, sparking controversy by requiring developer source code access. This highlights their diverging approaches to AI integration. Alipay is testing "Project Treasure," an optional AI-native interface replacing traditional app grids with a conversational window. Users can command complex tasks (e.g., "book a ride and order coffee") handled end-to-end by AI. This shift follows an abandoned standalone AI app, focusing instead on enhancing its existing user base. For unmodified mini-programs, Alipay's AI uses "screen-reading" to simulate user interactions, bypassing the need for developer overhaul. It also introduced "Token Pay" for micro-transactions and "AI Wallets" for autonomous agent spending. WeChat, prioritizing its core social function, is taking an embedded approach. Its AI agent will operate within existing contexts like group chats and official accounts, assisting without a separate interface. To enable this, WeChat offers developers two paths: granting source code access for direct AI control ("Automatic Mode") or manually encapsulating services into standardized "Skills." Both place significant burden on developers. Key differences emerge in handling legacy services: WeChat demands developer cooperation (code or labor), while Alipay's screen-reading offers immediate, if potentially less stable, compatibility. Alipay's 3 billion AI transactions demonstrate user acceptance of AI-driven commercial actions. The divergent strategies may reshape mini-program ecosystems—Alipay passively "AI-fying" services, WeChat potentially favoring resource-rich developers—and set competing technical standards. Ultimately, the competition centers on where users entrust the command to "help me get things done."

marsbit2h ago

Fully Entering the AI Era: Alipay Bets on Conversation, WeChat Holds Fast to Social

marsbit2h ago

Trading

Spot
Futures
活动图片