Anthropic’s newly introduced, opt-in feature guarantees that vulnerability reports will be delivered by their most powerful artificial intelligence systems, such as Claude Mythos.
A fresh Anthropic initiative granting open-source projects access to security findings from advanced artificial intelligence systems has drawn applications from numerous cryptocurrency enterprises, including Ethereum client builder Nethermind alongside Bitcoin and Lightning wallet ZEUS.
Enrollment applications for the OSS Scanner were submitted by the projects on Friday; this tool had been rolled out by Anthropic a day prior to help open-source builders patch security flaws before malicious actors can exploit them.
“These reports will be generated by our strongest models (including Claude Mythos) to give open-source projects the largest defensive advantage,”
Anthropic said in its announcement.
These applications are being driven by blockchain firms pushing to leverage cutting-edge artificial intelligence systems to guard against increasingly sophisticated hackers. Security specialists have cautioned that uneven access might place defenders at a disadvantage while potent alternatives grow more widely accessible.
Building upon its previous efforts with Project Glasswing, the OSS Scanner was described by Anthropic as an opt-in offering. The firm explained that while open-source software is already routinely checked for flaws and reports are issued following human evaluation, this manual review procedure is slow, meaning that vulnerability details cannot always be shared as quickly as desired.
Vulnerability reports will be sent directly to participating projects by the OSS Scanner as soon as their code is checked.
Submissions within the OSS Scanner GitHub repository indicate that Nethermind has requested audits for its entire code base, whereas ZEUS—a self-custodial Bitcoin and Lightning wallet—has asked for its application to be inspected for flaws impacting transactions, private keys, and Lightning connectivity. VirtEngine, a decentralized cloud computing marketplace built as a Cosmos SDK chain, is listed as another applicant.
Additional applicants are comprised of creators of artificial intelligence assistants, agent-protection utilities, and machine learning backbones, alongside ventures delivering software engineering instruments, cloud data storage, and energy grid management systems.
None of the merge requests had been processed as of publication time.
Evaluation of the projects will be handled by Anthropic on an individual basis, taking into account infrastructural importance, user protection levels, vulnerability to remote exploits, and reliance from users or other platforms.
Crypto Firms Face Growing Threats From AI-Assisted Attacks
AI-assisted breaches have been documented by cryptocurrency companies throughout the year. Operations were paused by Bitcoin swap provider Boltz in August because exploits were being developed by malicious actors quicker than the team could patch them, whereas crypto-payment processor PayPerQ disclosed recurring assaults suspected to be driven by artificial intelligence.
On Thursday, Anthropic cautioned that artificial intelligence might benefit malicious actors in the near future because exploitation is getting cheaper, whereas checking and repairing flaws remains slow and reliant on human effort.
