Back
Podcast

Decoding 2025's biggest web3 hacks: lessons & trends

This X Spaces discussion hosted by QuillAudits brought together leading security experts from across the Web3 ecosystem, including Guardrail Founder, Samridh Saluja, to discuss the alarming trend of security breaches in early 2025. The conversation focused on the three major attack vectors responsible for 95% of funds lost: access control issues, social engineering attacks, and infrastructure vulnerabilities.

TL;DR:

The discussion highlighted that while technical vulnerabilities remain important, the human element through social engineering and access control represents the dominant threat in 2025. Success requires a holistic approach combining technical controls, operational security, real-time monitoring, and most importantly, a security-first culture that scales with the value being protected.

"Everyone has a plan until you get punched in the face."
Samridh Saluja, Founder at Guardrail

Key stats discussed:

  • H1 2025 losses already exceeded total 2024 losses ($2+ billion)
  • Access control issues: 70% of funds lost ($1.6 billion)
  • Social engineering: 15% of funds lost
  • Three attack vectors combined: 95% of total losses

Listen to the X Spaces recording:

Main discussion points:

1. Social Engineering Attacks

The panel agreed that social engineering attacks are becoming increasingly sophisticated with AI tools, making them harder to detect. Key takeaways:

  • Good OPSEC is essential but must be tailored to risk levels
  • Isolated devices for crypto transactions recommended
  • Pig butchering scams targeting new crypto users
  • AI advancement making attacks more convincing

2. Access Control Issues

The dominant attack vector discussed, encompassing everything from private key compromises to infrastructure vulnerabilities:

  • Not just smart contract issues - includes Web2 infrastructure
  • Defense in depth approach needed
  • Role-based access control crucial but complex to implement
  • Regular monitoring and time locks recommended
  • Proxy-in-the-middle attacks emerging as new threat

3. Protocol-Specific Security Measures

Each speaker shared unique approaches:

  • EigenLayer's withdrawal queue system prevents immediate fund extraction
  • ZKsync's enhanced monitoring and video verification for access requests
  • Guardrail's real-time monitoring across 45+ chains

Key security recommendations:

  1. Implement defense in depth - Multiple layers of security
  2. Scale security investment with value - More value = more security
  3. Focus on permanent state changes - Protect irreversible actions
  4. Practice incident response - Quarterly drills minimum
  5. Use real-time monitoring - Detect attacks before completion
  6. Verify human requests - Video calls for sensitive access
  7. Isolate crypto operations - Dedicated devices and networks
  8. Stay updated on AI threats - Attackers using latest tools

Notable quotes:

  • "Good OPSEC cannot be really defined for every entity by a single entity. It basically depends on the level of risk that you're operating at." - Raul
  • "Everyone has a plan until you get punched in the face." - Samridh
  • "When security is working well, you don't want to hear about it. You only hear about it when something's going wrong." - Samridh
  • "The first focus that you should have...is on the place in your system where a permanent action is happening." - Raul

Speaker timestamps & key contributions:

Opening & Introduction

[00:00:00] @0xnoveleader (Quill Audits)

  • Welcome and introduction of speakers
  • Overview of H1 2025 hack statistics
  • Introduction of three major attack vectors

Social Engineering Discussion

[00:03:00] @0xnoveleader (Quill Audits)

  • Introduces social engineering topic
  • Notes 15% of funds lost to these attacks

[00:04:00] @SamridhSaluja (Guardrail)

  • AI making social engineering worse
  • Importance of integrity monitoring
  • Well-funded adversaries growing stronger

[00:08:00] @saxenism (ZKsync)

  • Good OPSEC depends on risk level
  • Comparison to testing strategies
  • Need for layered security approach

[00:11:00] @blocksek (EigenCloud)

  • EigenLayer's withdrawal queue mechanism
  • Allows intervention even after compromise
  • Prevents withdrawals to different wallets

Access Control Deep Dive

[00:13:00] @0xnoveleader (Quill Audits)

  • 70% of funds lost to access control issues
  • Links to social engineering and infrastructure

[00:14:00] @saxenism (ZKsync)

  • Bybit hack discussion
  • Web2 vs Web3 access control
  • Implementation of comprehensive monitoring

[00:17:00] @SamridhSaluja (Guardrail)

  • AI tools making attacks easier
  • Private key compromise detection strategies
  • Multisig monitoring importance

[00:20:00] @blocksek (EigenCloud)

  • Proxy-in-the-middle attacks
  • Smart contract deployment vulnerabilities
  • Best practices for proxy initialization

[00:22:00] @officer_cia (Remedy)

  • Importance of understanding access privileges
  • Segregating roles effectively
  • Impact awareness for access holders

Protocol-Specific Insights

[00:25:00] @blocksek (EigenCloud)

  • Introduction to EigenCloud evolution
  • Verifiable cloud infrastructure
  • Suite of verification mechanisms
  • Security considerations for different verification types

[00:35:00] Samridh Saluja (Guardrail)

  • Guardrail platform overview
  • Real-time monitoring across chains
  • Common response failures from teams
  • "It won't happen to us" fallacy

[00:45:00] @saxenism (ZKsync)

  • Zero-knowledge proof security considerations
  • Don't optimize ZK code
  • Use battle-tested implementations
  • Physical security for remote teams
  • Video verification for access requests

Career Advice & Closing

[00:53:00] @blocksek (EigenCloud)

  • Recommends Cipher Updraft for learning
  • Participate in competitions
  • Specialize in emerging tech like ZK
  • Balance AI usage with fundamental knowledge

[00:56:00] @0xnoveleader (Quill Audits)

  • Closing remarks
  • Thanks to all speakers