AI Briefing
GeekNews news
The latest AI and developer news from GeekNews, with short summaries.
Visit website →Codex CLI adds /goal feature
Codex CLI 0.128.0 adds `/goal`, a goal-based automatic iteration execution feature.
DeepClaude - Cutting Claude Code Agent Loop Costs 17x Using DeepSeek V4 Pro
A tool that replaces Claude Code CLI's backend with DeepSeek and others to cut agent costs by up to 17x.
Long-Running Agents - What Changes When Agents Run for Days
An analysis of the architecture and key challenges required when AI agents run autonomously for days beyond a single session.
OpenAI o1 correctly diagnosed 67% of ER patients while triage doctors scored 50-55%
OpenAI o1 outperformed doctors in emergency room note-based diagnosis and treatment planning.
Kimi K2.6 Beats Claude, GPT-5.5, and Gemini in Coding Challenge
Moonshot AI's Kimi K2.6 won an AI coding challenge, edging out GPT-5.5 and Claude.
Russia is polluting Wikipedia
Russia is attempting information manipulation by spreading pro-Kremlin narratives targeting Wikipedia and AI tools.
Show GN: CTX: Memory Between Claude Code Sessions — Install via pip install or /plugin install
CTX gives Claude Code local session memory and retrieval-based context injection.
Building a Web of Trust to Fight LLM Spam
Tangled has introduced vouching trust signals to counter LLM spam.
California starts issuing tickets to autonomous vehicles that violate traffic laws
Starting July 1, California will implement a system that directly notifies manufacturers when autonomous vehicles commit violations.
AI Self-Preference in Algorithmic Hiring: Empirical Evidence and Implications
Hiring bias was confirmed in which LLMs favor resumes they themselves generated.
VS Code Inserts 'Co-Authored-by Copilot' into Commits Regardless of Usage
A regression has surfaced where VS Code adds `Co-authored-by: Copilot` regardless of whether Copilot was actually used.
Context Mode - An MCP Server That Cuts AI Coding Agents' Context Window by 98%
Context Mode, built for AI coding agents, has been introduced as reducing context usage by 98%.
DeepSeek V4 - Nearly Frontier-Level, at a Much Lower Price
DeepSeek released previews of V4-Pro and V4-Flash, emphasizing price competitiveness.
Uber Burned Through Its Entire 2026 AI Budget in Just 4 Months with Claude Code
Uber's rapid adoption of Claude Code led it to exhaust its 2026 AI budget in just 4 months.
Apple shipped a Claude.md file bundled with the Support app
Claude.md was included in Apple Support app v5.13 but was removed in 5.13.1.
Gay jailbreak technique
Introduced an LLM jailbreak technique that mixes LGBT identity and role-play.
Fincept Terminal - Open-Source Financial Analysis Platform
An open-source financial analysis platform supporting 37 AI agents and multiple LLMs has been released.
Spotify added a 'Verified' badge to distinguish human artists from AI
Spotify has introduced a 'Verified by Spotify' badge to identify human artists.
AI Uses Less Water Than the Public Thinks
Annual water use by California AI data centers is estimated at 0.055% of total human water use.
Claude Opus 4.7 Really Does Know Kelsey
Claude Opus 4.7 identifies a specific author from short pieces of writing across various genres, hinting at a threat to anonymity in the AI era.