News linked to both this project and an event.
OpenAI has released the Frontier Governance Framework, systematically elaborating on how its AI safety and governance practices align with emerging regulatory requirements such as the California Frontier AI Transparency Act and the EU's General-Purpose AI Code of Conduct. Based on OpenAI's existing Preparedness Framework, this framework focuses on areas including cyberattacks, CBRN risks, harmful manipulation, loss of control risks, model reporting, security incident response, and external expert review. It also states that it will be continuously updated as model capabilities and the regulatory environment evolve.
According to Decrypt, OpenAI CEO Sam Altman stated that Anthropic is promoting its AI model Claude Mythos through “fear-based marketing,” using narratives about security risks to justify its limited-open strategy. Claude Mythos has recently drawn attention for its ability to autonomously discover software vulnerabilities and perform complex cybersecurity operations. The report notes that Mozilla previously disclosed that the model identified 271 vulnerabilities in the Firefox browser during testing. Meanwhile, discussions surrounding the model’s potential offensive cybersecurity risks continue to intensify. Altman also emphasized that OpenAI will not scale back its infrastructure investments and will continue expanding its computational capabilities.
Ledger Chief Technology Officer Charles Guillemet pointed out that the development of post-quantum cryptography has entered a critical stage. Although the timeline for a practical quantum computer remains unclear, a full-scale migration of the encryption systems across the industry is an inevitable trend. Led by NIST, the traditional sector plans to phase out high-risk algorithms by 2030 and completely ban them by 2035, with government and enterprise institutions expected to complete their migration layouts by 2029. Encryption and key exchange will adopt ML-KEM to defend against quantum decryption attacks on harvested data, with digital signatures becoming the core of blockchain transformation. The traditional industry prefers ML-DSA hybrid schemes, while the blockchain sector favors the more secure and robust SLH-DSA hash-based signature. Both schemes have their respective advantages and disadvantages. The compatibility challenges of post-quantum algorithms with MPC and threshold signatures remain a key risk that the industry urgently needs to address.