BREAKING: $435 million has been liquidated from the cryptocurrency market in the past 12 hours.

On May 1, OpenAI unveiled the latest demo video for Codex, highlighting its AI coding and office workflow capabilities. The official statement noted that users can view real-time progress, access to files and tools, and follow-up steps while Codex executes tasks. The demo also showed that users can directly open generated files in the same thread, submit edit requests, and iteratively collaborate on content—creating a smooth workflow that moves from draft to presentation deck. OpenAI noted that Codex is designed to help users "get work done faster."

On May 1st, Anthropic’s Claude announced that its code security tool, Claude Security, has launched public beta testing for Claude Enterprise users. Per the official announcement, Claude Security can scan code repositories for security vulnerabilities, conduct secondary validation of detection results to reduce false positives, and automatically generate remediation solutions for developers to review and approve. Anthropic noted the tool operates without requiring API integration or self-built agents—leveraging Claude Opus 4.7 directly. Anthropic revealed that since its research preview launched in February this year, hundreds of organizations have used Claude Security in production environments and uncovered issues missed by traditional scanners. Based on user feedback, the team has added new features including scheduled scanning, directory-level scanning, CSV/Markdown exports, and Webhook notifications.

Cursor announced on May 1 that its Security Review feature is now available to Teams and Enterprise users. Per the official announcement, the feature includes two persistent AI security agents: the Security Reviewer (which automatically audits security vulnerabilities in every PR and leaves feedback) and the Vulnerability Scanner (which regularly scans the full codebase and shares findings via Slack). Cursor also offers team-customizable security agents, enabling teams to tweak trigger conditions, add custom commands, integrate proprietary tools, and set up output sharing preferences. The Cursor team is continuously refining the runtime, testing framework, and model capabilities powering the Security Review to boost the default security review experience.

On May 1, sources in the developer community report that xAI has officially launched the Grok 4.3 API. From released visuals, Grok 4.3 is now listed in xAI’s API model catalog, supporting a context window of up to 1 million tokens and offering text input/output, multimodal, and tool-calling functionality. xAI also rolled out the grok-4.20 multi-agent inference model, along with various configurations including fast reasoning and non-reasoning variants. Pricing for the Grok 4.3 API is set at $1.25 per million input tokens and $2.50 per million output tokens.

On May 1, ahead of the Drift Protocol attack, North Korean hackers spent months conducting offline outreach and infiltrating company employees—ultimately pulling off one of the crypto industry’s largest-ever social engineering attacks, which resulted in roughly $285 million in losses. Blockchain security firm TRM Labs’ data indicates that as of 2026, North Korean hackers have accounted for 76% of all cryptocurrency hacking losses.

On May 1, Iranian Foreign Ministry Spokesperson Baghaei said a quick negotiated settlement between Iran and the United States is unrealistic. He noted Iran has deep mistrust and suspicion of U.S. actions, accumulated over 47 years since the Islamic Revolution’s victory. He emphasized results won’t come quickly, stating: “After all, this is an extremely bloody conflict.” Baghaei labeled the U.S. and Israel “extremely cruel,” citing the Gaza Strip and Lebanon conflicts over the past two and a half years. “Whoever the mediator is, an agreement in the short term is unrealistic under these circumstances,” he added. (CCTV)