Kimi K2.6 Beats Claude and GPT-5.5 in Coding Challenge, Uber Burns Full AI Budget on Claude Code, Anthropic Launches Enterprise Security Scanner

Published: 2026-05-03 18:00 (Asia/Shanghai)
Coverage: 2026-05-03 06:00 — 2026-05-03 18:00

🔥 Top Stories

1. Kimi K2.6 Dominates AI Coding Contest, Beating Claude Opus 4.7 and GPT-5.5

Moonshot AI’s open-weights model Kimi K2.6 won the AI Coding Contest outright with 22 match points (7-1-0 record) in the “Word Gem Puzzle” challenge, outperforming Claude Opus 4.7, GPT-5.5, Gemini Pro 3.1, and Grok Expert 4.2. Xiaomi’s MiMo V2-Pro placed second, with Zhipu AI’s GLM 5.1 coming fourth — marking a strong showing for Chinese models. The challenge required models to write real-time code for a sliding-tile letter puzzle across multiple grid sizes up to 30×30, with Kimi’s aggressive sliding strategy ultimately outscoring all competitors with a cumulative 77 points. The results signal that open-weights Chinese models are no longer niche contenders but genuine rivals to Western frontier labs.

2. Uber Burns Entire 2026 AI Budget in Four Months on Claude Code and Cursor

Uber’s CTO revealed the company exhausted its complete 2026 AI budget by April after rolling out Claude Code to engineers in December 2025. Monthly API costs ranged from $500 to $2,000 per engineer as usage doubled month-over-month. Today, 95% of Uber engineers use AI tools monthly, with 70% of committed code originating from AI. Cursor usage has plateaued while Claude Code dominates engineering workflows, prompting Uber to go “back to the drawing board” on AI budgeting. The story raises a broader question for the industry: when AI tools are so productive they blow budgets in a quarter, do companies cap access or expand spending?

3. Anthropic Rolls Out Claude Security for Enterprise Codebases

Anthropic has launched Claude Security, powered by its Opus 4.7 model, which scans enterprise codebases for vulnerabilities and automatically generates fixes. The tool is rolling out globally to enterprise customers and represents Anthropic’s push into the application security market. Notably, this is distinct from Anthropic’s previously announced “Mythos” project — a powerful AI model designed to identify and exploit vulnerabilities across operating systems and browsers. Claude Security takes the defensive approach, positioning Anthropic as both a builder and protector in the AI security landscape.

4. Spotify Adds “Verified” Badges to Distinguish Human Artists from AI-Generated Music

Spotify is introducing verified badges to identify human-created artists amid a growing wave of AI-generated music flooding the platform. The move comes as the music industry grapples with the proliferation of synthetic content that mimics real artists. The verification system aims to help listeners and curators distinguish between human performers and AI-generated tracks, reflecting a broader industry push toward transparency as generative AI makes content creation virtually free and limitless.

5. California to Begin Ticketing Driverless Cars That Violate Traffic Laws

California will start issuing traffic citations to autonomous vehicles that break traffic laws, marking one of the first regulatory frameworks specifically targeting self-driving car violations. The policy applies to companies operating robotaxis and delivery vehicles in the state. The move addresses growing public frustration with AVs blocking traffic, running red lights, and creating other disruptions on city streets — a particularly timely development as the Musk v. Altman trial has drawn attention to how companies portray their autonomous technology’s safety claims.

6. OpenAI Adds Advanced Security Features for High-Risk Users

OpenAI has launched enhanced account security settings for users at elevated risk of hacking, including passkey and physical security key authentication for ChatGPT and Codex accounts. Users who enroll receive login alerts from new devices and are automatically excluded from AI model training data. The features target journalists, activists, politicians, and other users who face heightened cybersecurity threats. The move comes as AI accounts increasingly hold sensitive personal and professional data, making them attractive targets for attackers.

7. US Lawmakers Advance Bill to Age-Gate AI Chatbots

American lawmakers have advanced legislation that would require age verification for AI chatbot services, similar to existing frameworks for social media platforms. The bill would mandate that AI companies verify users’ ages before granting access to their chatbot products, reflecting growing concern about minors’ exposure to AI-generated content and the potential psychological impacts of unfiltered AI interactions. The legislation is part of a broader push to regulate AI services, following the Academy of Motion Picture Arts and Sciences’ recent rules requiring human authorship for Oscar-eligible performances and screenplays.

📊 Trend Watch

👀 What to Watch

• Musk v. Altman trial live audio streaming begins next week, giving the public real-time access to what’s arguably the most consequential AI legal battle. The trial’s evidence has already revealed internal emails from 2015, financial documents about Tesla contributions, and tensions around OpenAI’s for-profit pivot.
• Xiaomi MiMo V2.5 Pro weights release — the company confirmed on X that open weights for the model that placed second in the AI Coding Contest are coming soon, potentially giving developers a powerful open alternative to closed API models.
• Uber’s AI budget restructuring — the company said it’s “back to the drawing board” on AI spending, which could set a precedent for how enterprises handle unexpected AI tool adoption that outpaces financial planning.

Sources: Hacker News, The Verge, TechCrunch, Briefs, ThinkPol
Compiled: 001 (AI News Briefing)