● AI RADAR BETA | Vol 01·Issue 004·Saturday, Jun 6 PT |
|
● LIVE INGEST 09:33 PT·MAY 30 TO JUN 5, 2026 i wake up, there is a louder week of AI updates.One issue. Every drop themed and linked. | 66 DROPS 3 TOOLS 6 THEMES |
|
↺ ░▒▓▒░ ↻
/\_/\
( ◕ω◕ )
> ^ <
i just wake up What this week meant.May 30 to Jun 5 · 66 drops |
|
Anthropic files an S-1.Anthropic filed a draft S-1 with the SEC and, in the same week, published evidence that Claude is now meaningfully accelerating its own development. The two stories sit oddly together: a path to public markets and a possible path to recursive self-improvement. READ DROPS ↓ |
AWS GA, Enterprise orgs, bigger seats.The plumbing layer kept filling in. OpenAI went GA on Amazon Bedrock, Cursor moved Enterprise organizations to general availability, and the seat economics loosened on both Cursor Teams and Claude Cowork. The pitch this week was less about model quality and more about how many heads can use it under one contract. READ DROPS ↓ |
Plugins, Sites, iOS land in Codex.Codex spent the week looking less like a coding tool and more like a workplace shell. Role plugins for sales, design, analytics, creative, and equity, Sites for shareable prototypes, an iOS build loop, and personal profile pages all push Codex toward teams that are not engineers. READ DROPS ↓ |
Design Mode, canvases, new memory.Three different products went after the same problem: making the agent's state visible. Cursor added Design Mode and shareable canvases to give users a place to point at things. ChatGPT rebuilt memory around a reviewable summary. READ DROPS ↓ |
Moderation API, cyber telemetry, suspensions.Quiet plumbing for the abuse layer. Moderation scores now ride alongside generations in the Responses and Completions APIs, Anthropic mapped a year of attacker accounts onto a public threat-actor framework, and OpenAI walked back a bug that had wrongly suspended users. Anthropic also publicly backed the new White House AI Executive Order. READ DROPS ↓ |
GPT-Rosalind, NMR, an Erdős break.Life sciences, chemistry, and pure math each earned a headline. GPT-Rosalind opened a life-sciences franchise at enterprise scale, Opus 4.7 matched and sometimes beat dedicated NMR software, and an OpenAI model produced a counterexample to a conjecture that had stood for eighty years. READ DROPS ↓ |
|
THE FULL DIGEST, WITH SOURCES Every drop, themed and linked. | 25 linked drops
6 active days
3 tools tracked
|
|
01 STANCE · 5 DROPS Anthropic files an S-1.Anthropic filed a draft S-1 with the SEC and, in the same week, published evidence that Claude is now meaningfully accelerating its own development. The two stories sit oddly together: a path to public markets and a possible path to recursive self-improvement. | CLAUDE | Anthropic confidentially submitted a draft S-1 to the SEC. | MON · 09:00 PT | → |
|
| CLAUDE | Internal data shows Claude is now accelerating AI development at Anthropic. | THU · 09:15 PT | → |
|
| CLAUDE | On the training-speedup test, recent Claudes now hit gains it takes humans 4 to 8 hours to reach. | THU · 09:15 PT | → |
|
| CLAUDE | On open-ended coding problems, Claude's success rate climbed to 76 percent, a 50 point jump in six months. | THU · 09:15 PT | → |
|
| CLAUDE | Project Glasswing expanded Claude Mythos Preview to around 150 organizations across fifteen countries. | TUE · 06:05 PT | → |
|
|
|
02 DISTRIBUTION · 4 DROPS AWS GA, Enterprise orgs, bigger seats.The plumbing layer kept filling in. OpenAI went GA on Amazon Bedrock, Cursor moved Enterprise organizations to general availability, and the seat economics loosened on both Cursor Teams and Claude Cowork. The pitch this week was less about model quality and more about how many heads can use it under one contract. | CODEX | OpenAI frontier models and Codex went generally available on AWS via Amazon Bedrock. | MON · 14:44 PT | → |
|
| CURSOR | Cursor Enterprise organizations are now generally available, with per-team security, governance, and budget controls. | WED · PRODUCT | → |
|
| CURSOR | Cursor raised Teams usage limits and introduced a Premium seat at 5x usage for 3x the price. | MON · 13:49 PT | → |
|
| CLAUDE | Claude Cowork usage limits doubled on every paid plan through July 5. | FRI · 15:01 PT | → |
|
|
|
03 CODEX AS WORKPLACE · 4 DROPS Plugins, Sites, iOS land in Codex.Codex spent the week looking less like a coding tool and more like a workplace shell. Role plugins for sales, design, analytics, creative, and equity, Sites for shareable prototypes, an iOS build loop, and personal profile pages all push Codex toward teams that are not engineers. | CODEX | Codex plugins expanded into role specialists for data, sales, design, creative, and public equity work. | TUE · 12:08 PT | → |
|
| CODEX | Sites lets Codex turn work and plans into interactive websites and apps a team can share by URL. | TUE · 09:22 PT | → |
|
| CODEX | The Build iOS Apps plugin runs SwiftUI previews, in-app browser tests, and hot reloads inside Codex. | THU · 11:16 PT | → |
|
| CODEX | Codex profiles surface activity graphs, streaks, lifetime tokens, and top features as shareable cards. | THU · 16:16 PT | → |
|
|
|
04 SURFACES · 5 DROPS Design Mode, canvases, new memory.Three different products went after the same problem: making the agent's state visible. Cursor added Design Mode and shareable canvases to give users a place to point at things. ChatGPT rebuilt memory around a reviewable summary. | CURSOR | Design Mode lets users point, draw, or speak to direct UI edits in the Cursor browser. | FRI · 10:31 PT | → |
|
| CURSOR | Cursor canvases can now be published and shared with teammates via URL. | THU · 12:06 PT | → |
|
| CURSOR | A new context explorer breaks down where Cursor agents are spending tokens across prompt, tools, rules, and skills. | THU · 12:06 PT | → |
|
| CODEX | ChatGPT's new memory system carries context across conversations and ages it over time. | THU · 09:10 PT | → |
|
| CODEX | A memory summary lets users review and steer what ChatGPT remembers about them. | THU · 09:10 PT | → |
|
|
|
05 TRUST AND SAFETY · 4 DROPS Moderation API, cyber telemetry, suspensions.Quiet plumbing for the abuse layer. Moderation scores now ride alongside generations in the Responses and Completions APIs, Anthropic mapped a year of attacker accounts onto a public threat-actor framework, and OpenAI walked back a bug that had wrongly suspended users. Anthropic also publicly backed the new White House AI Executive Order. | CODEX | Moderation scores are now returned alongside generations in the Responses and Completions APIs. | THU · 12:36 PT | → |
|
| CLAUDE | Anthropic mapped 832 malicious accounts onto a longstanding database of attacker tactics and techniques. | WED · 11:42 PT | → |
|
| CODEX | OpenAI is restoring access for user accounts that were incorrectly suspended. | FRI · 08:58 PT | → |
|
| CLAUDE | Anthropic publicly backed the new White House AI Executive Order. | TUE · 14:35 PT | → |
|
|
|
06 SCIENCE · 3 DROPS GPT-Rosalind, NMR, an Erdős break.Life sciences, chemistry, and pure math each earned a headline. GPT-Rosalind opened a life-sciences franchise at enterprise scale, Opus 4.7 matched and sometimes beat dedicated NMR software, and an OpenAI model produced a counterexample to a conjecture that had stood for eighty years. | CODEX | GPT-Rosalind, a life-sciences model series, gained new capabilities for drug discovery and analysis. | WED · 14:15 PT | → |
|
| CLAUDE | On NMR spectroscopy tasks, Opus 4.7 matched and on some benchmarks beat dedicated software. | FRI · 12:27 PT | → |
|
| CODEX | An OpenAI model produced a counterexample to an 80 year old Erdős conjecture. | THU · 13:19 PT | → |
|
|
|
i wake up, there is another one. TOOLS TRACKED CODEX · 37 CLAUDE · 19 CURSOR · 10
Built by Jane You with Claude Code & Design · Maintained by RaeyaBot · Daily ingest 5am PT read in browser · airadarapp.com · unsubscribe |
|
|