AI Benchmarks, Code Agents & IP Law
OpenAI Debuts LifeVal and Qodo Plugin as Publishers Sue Brave
OpenAI benchmarks everyday tasks for 1B weekly ChatGPT users while Qodo pairs GPT-6 Astra with Codex, Anthropic updates Opus 5.5, and newspapers sue Brave.

Your listening Queue & catch up
Add a story or build a catch-up queue to start listening.
Enterprise artificial intelligence developers unveiled new practical-task evaluations, multi-repository code review workflows, and copyright guardrails on Monday, even as major news publishers escalated federal litigation over web scraping.
OpenAI Launches LifeVal Benchmark for ChatGPT's 1 Billion+ Weekly Users
In a research post published Monday by OpenAI News, OpenAI introduced LifeVal, an internal and public evaluation suite designed to measure how effectively ChatGPT—now used by more than 1 billion people every week—handles multi-step personal assistant tasks across browser use, computer use, and software connectors:
- Time-Use Datasets & 3,995-Respondent Survey: To identify high-value daily workflows across caregiving, health, education, household administration, personal finance, shopping, and government services, OpenAI combined the American Time Use Survey and the Multinational Time Use Study with a demographically representative survey of 3,995 respondents.
- Practical Task Scores: On sample LifeVal workflows, OpenAI's latest models scored 100% on locating a 1-family home with a separate living unit in Luzerne County, calculating the transfer tax, and adding the deed to a county records cart; 98.0% on organizing a flood-preparation volunteer action plan for 4 neighbors; 96.3% on configuring a remote game-watching stream with a backup link; 93.1% on narrowing local kindergartens to 3 finalists with 2 open family preferences; and 90.4% on preparing a 1-week care and meal plan for a parent adjusting to a new diabetes diagnosis.
Qodo Integrates GPT-6 Astra and Decisions API Into Codex Code Review Plugin
In a companion engineering profile on OpenAI News, AI code-quality platform Qodo detailed how its new Codex marketplace plugin uses GPT-6 Astra to prevent production regressions across enterprise codebases spanning 10,000 or more repositories:
- Pre-PR Cross-Repository Context: Qodo co-founder and CEO Itamar Friedman and AI Developer Relations Lead Nnenna Ndukwe explained that while Codex excels at local coding tasks, Qodo's Agentic Toolbox feeds organizational rules, pull-request history, and cross-repository dependency graphs—including services linked via Kafka across multiple Git providers—into the same session before a developer opens a pull request. Internally, Ndukwe runs GPT-6 Astra at Extra High and Ultra effort settings to automate weekly engineering reviews and skill refinement.
- Selecting From 200 Rules via the Decisions API: Friedman noted that Qodo plans to use OpenAI's Decisions API to select which of 200 enterprise coding rules apply at a given stage of development before invoking a reasoning model, cutting latency and inference cost while surfacing only unresolved disagreements between coding and review agents for human judgment.
Anthropic Updates Claude Code on Opus 5.5; Newspaper Publishers Sue Brave Software
As coding and search agents proliferate, intellectual property disputes intensified across both gaming and journalism on Monday:
- Anthropic Blocks Video Game Clones in Claude Code: According to The Verge senior reporter Jay Peters (citing an initial report by Tom's Hardware), Anthropic updated Claude Code's system prompt on Opus 5.5 after retro gaming communities saw a wave of AI-generated decompilations and ports. Anthropic's Addy Osmani confirmed that the company added 2 sentences to Claude Code's prompt on Opus 5.5 to steer the model away from recreating existing commercial titles—which violates Anthropic's IP usage policy—while preserving support for original games, genre games, and projects built from a user's own designs and assets.
- New York Times and Tribune Outlets Sue Brave Over 30B-Page Index: Meanwhile, TechCrunch reported that The New York Times—joined in a separate Monday complaint by the Daily News, the Chicago Tribune, and the Orlando Sentinel—sued 11-year-old browser developer Brave Software (founded in 2015, with roughly $250 million raised) in federal court in Manhattan. The lawsuits allege that Brave used an unidentified web crawler to compile an index of more than 30 billion web pages, licensed scraped publisher articles to AI companies including Anthropic, Perplexity, Mistral, xAI, and Cohere, and offered browser tools that bypass publisher paywalls and block publisher advertisements.
Sources
Follow the coverage
Published reporting connected by OpenAI.
- OpenAI Unifies ChatGPT Work & Codex Limits; Binance $3.64B Outflow
- OpenAI Plans Campaign to Rebrand Data Centers as ‘AI Compute Campuses’
- OpenAI Debuts Grid Program as Omen AI Buys Aquatic Labs for $16M
- OpenAI Debuts LifeVal and Qodo Plugin as Publishers Sue BraveYou are reading this story
Comments (0)
No comments yet. Be the first to share a thought on this story.
Request a correctionContact the newsroom