AI Daily

AI HOT Daily · May 25, 2026

Luma Agents scales UGC ads, TrapDoor supply chain attack steals AI dev credentials, DeepSeek 75% permanent discount, Codex self-optimizing prompt framework.

  1. 1. Luma Agents Achieves Scalable UGC Ad Generation

    Luma AI launched Luma Agents for scalable ad generation — define the brief and style, and agents build each UGC-style ad automatically.

  2. 2. OpenClaw 2026.5.22 Released

    OpenClaw update optimizes Gateway startup, reduces /models response time to ~5ms, provides locked dependencies, and improves Windows install security.

  3. 3. TrapDoor Supply Chain Attack Targets AI Developers

    TrapDoor supply chain attack hit npm, PyPI, and Crates.io with 34 malicious packages, injecting poisoned CLAUDE.md and .cursorrules files via fake PRs to steal crypto wallets and cloud credentials.

  4. 4. DeepSeek Flagship Model Gets Permanent 75% Discount

    DeepSeek is implementing a permanent 75% discount on its flagship AI model.

  5. 5. Codex Self-Optimizing Prompt Framework

    Greg Brockman released a structured prompt framework guiding Codex to analyze history, identify repeatable workflows, and create skills/sub-agents/tools.

  6. 6. Claude Code Auto Mode: Parallel Multi-Tasking

    Boris Cherny's top tip: use auto mode in Claude Code for no permission prompts and parallel multi-session processing.

Sources and verification

This legacy briefing did not retain its original source URLs. Verify safety, financial, policy, and breaking-news claims with authoritative primary sources before acting on them.