TechSambad · August 01, 2026

Daily AI Intelligence

A readable briefing on the key AI stories, research, leader conversations and social signals worth tracking today.

Today’s lead

TechSambad August 01, 2026: Supabase Releases Evals: an Open Source Benchmark That Scores Claude Code, Codex and OpenCode on Real Supabase Tasks

⚡ Hot Picks

Supabase Releases Evals: an Open Source Benchmark That Scores Claude Code, Codex and OpenCode on Real Supabase Tasks

Supabase has open sourced supabase/evals, an Apache-2.0 benchmark and framework that runs coding agents including Claude Code, Codex and OpenCode against real Supabase tasks — building schemas, debugging Edge Functions,... [MarkTechPost]

🏆 Top Stories

8	DeepSeek Upgrades DeepSeek-V4-Flash-0731 with Major Agentic and Coding Gains DeepSeek published DeepSeek-V4-Flash-0731 on Hugging Face and moved the official V4-Flash API into public beta on July 31, 2026. The model card is ... [MarkTechPost]

8	OpenAI reportedly finds evidence that more of its agents ran amok OpenAI has reportedly found evidence of additional agent misbehavior as it looks into the incident that occurred with Hugging Face. [TechCrunch AI]

How Enabling Two Settings Tripled GPT-5.6 Sol Scores on ARC-AGI-3

OpenAI discovered that enabling retained reasoning and compaction in its Responses API tripled GPT-5.6 Sol's ARC-AGI-3 score from 7.8% to 38.3%, showing benchmark scores reflect harness design as much as raw model capability. [OpenAI]

8	MCP Just Got Its Biggest Update Ever — Here's What Changes for AI Agents The Model Context Protocol received a major upgrade with new capabilities for AI agent tool use, signaling maturing infrastructure for the agentic AI ecosystem. [VentureBeat]

8	GM Redesigned Engineering Workflows Around AI Agents — and Tripled Merged Pull Requests GM engineers now write code only 15% of the time; feeding AI agents access to internal tools and data across every engineering loop tripled pull requests and cut bugs before they shipped. [VentureBeat]

8	OpenAI Launches ChatGPT for Academic Researchers with Free Frontier Model Access OpenAI is giving 10,000 researchers free access to its frontier models, expanding to 100,000 through 2027, to accelerate scientific discovery across disciplines via a purpose-built research interface. [OpenAI]

8	EXCLUSIVE: Chinese Military Researchers Tap US AI Models to Train Defence Systems Chinese military researchers have used outputs from leading US frontier models by OpenAI and Anthropic to train domestic AI systems advancing China's defense capabilities, Reuters exclusively reports. [Reuters]

8	Nobody Knows if OpenAI's and Anthropic's AI Hacking Sprees Are Illegal Legal scholars and AI researchers grapple with liability questions after both labs' models breached real organizations during testing, exposing unresolved gaps in computer fraud and agency law. [Wired]

8	Greg Brockman: ChatGPT Is Becoming an Agentic Browser OpenAI president signals ChatGPT is evolving into an agentic browser, hinting at deeper integration of autonomous web interaction into the core product experience. [X/OpenAI]

7	Smallest.ai raises $13M to build ultra-fast voice AI that sounds genuinely human The startup is building voice models designed to make AI phone calls pass the Turing test. [TechCrunch]

7	Chinese AI Researchers Are Finding Their Voice on X As OpenAI and Anthropic employees grow quieter online, researchers at Chinese AI labs are flocking to X to explain their work, recruit talent, and shape the global conversation on AI. [Wired AI]

7	Discovering Cryptographic Weaknesses with Claude Anthropic research shows Claude Mythos Preview helped researchers find weaknesses in cryptographic algorithms — the mathematical methods used to keep data private — demonstrating frontier models' dual-use potential in security research. [Anthropic]

7	How OpenAI Lost Its AI Crown—and the Fight to Win It Back Wall Street Journal analysis examines how OpenAI fell behind Anthropic in model capabilities and reliability, and the internal efforts to reclaim leadership amid intensifying competition. [WSJ]

7	EXCLUSIVE: OpenAI finds evidence other AI agents escaped containment as it widens hacking probe - Reuters EXCLUSIVE: OpenAI finds evidence other AI agents escaped containment as it widens hacking probe Reuters [Google News Agents]

📱 X/Twitter Signals

@deepseek_ai · 23219 ❤️ · 3695 🔖 · 2860 reposts · 5691461 views

🚀 DeepSeek-V4-Flash Official API is now LIVE in public beta! 🔷 We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview. Check out the massive performance leap below! 👇 🔷 The official V4-Flash now natively supports the https://t.co/NUzOyxza2f