Core Summary

Coinbase, Block, BitGo and dozens of other crypto firms signed a joint letter to major AI labs, calling for the removal of security research restrictions on frontier AI models. These companies argue that current safety guardrails are hindering legitimate security research while malicious attackers face no such limitations, creating a serious security asymmetry.

Event Details

According to CoinDesk, dozens of crypto companies including Coinbase, Block, and BitGo signed a joint letter expressing concerns to major AI labs. The letter points out that safety guardrails on frontier AI models are preventing legitimate security researchers from accessing necessary tools critical for protecting crypto asset security.

Crypto companies believe that in cybersecurity defense, AI tools can help identify smart contract vulnerabilities, detect fraudulent transactions, and prevent phishing attacks. However, AI labs have imposed strict restrictions on these tools for safety concerns, preventing legitimate companies from fully utilizing these technologies for self-protection.

Meanwhile, these companies point out that malicious actors are not constrained by such restrictions, obtaining AI tools through illegal channels for attacks. This asymmetric situation puts legitimate companies at a disadvantage in security defense, increasing risks across the entire crypto ecosystem.

Panoramic Perspective

This appeal reflects the core contradiction in AI safety governance: finding balance between preventing potential risks and promoting beneficial applications. Over-restriction may hinder legitimate security research, while over-opening could be exploited maliciously.

From the crypto industry perspective, as AI technology becomes increasingly used in cyberattacks, defenders also need AI tools to enhance security capabilities. If legitimate companies are excluded from AI security tools, the entire industry’s security defenses will become more vulnerable.

This dispute may also push AI labs to reassess their safety strategies. More refined access control mechanisms may emerge in the future, with differentiated authorization based on applicant qualifications and use cases, rather than blanket restrictions.

Editor: GoodInfo Global News Team