Miao "MJ" Jiang 🅝

The cybersecurity landscape has dramatically changed. Being able to build up your defenses using AI, while keeping that research private, is now critically important – and we want to make that available to everyone. We just rolled out NEAR AI Cybersecurity program: access to uncensored models for security scanning. All your code, prompts, and findings stay private. Everything runs inside secure enclaves and is encrypted end-to-end.

The funds from the $3.8M NEAR Intents exploit have been returned, just one day later, and the investigation is closed. Thanks to SHIELD, the AI security layer on Intents, along with some aggressive detective work, the Intents team identified the party responsible less than 24h after the hack, established communication, and got the funds back in full at 14:30 UTC today. Thank you to @alexauroradev and the whole @near_intents team for this fast and clean resolution – as well as the broader @NEARProtocol ecosystem and SHIELD partners for the support and help. There’s been relentless work over the past 24 hours by so many people to make this right and it is a great testament to this community’s values and determination. To echo Alex: for security researchers looking for exploits, we encourage you to use bug bounties. They exist for a reason. At NEAR, we believe that privacy is a right, but not as a way to facilitate crime. Our technology is built to expand human agency and opportunity, not to facilitate criminal activity. The events of the last week have shown the value of SHIELD to find suspicious activity before things escalate too far, as well as to accelerate resolution so the vast majority of legitimate users are not affected. Confidentiality cannot come at the expense of lawfulness and I strongly believe it doesn’t have to. We’ve learned a great deal from this incident and work is already underway to harden our security systems and further adapt to the changing cybersecurity landscape. AI may be accelerating exploits, but it can also help us defend against them. I hope our entire industry responds with vigilance and feels the same sense of urgency to work together.

Try confidential GLM-5.3-Flash on @near_ai

Completed: first quantum-safe signing scheme (ML-DSA) added to NEAR accounts. Up next: optimization & expanding support, securing post-quantum consensus, and updating cross-chain infra. Read how @NEARProtocol is securing the protocol & ecosystem for the post quantum future.