The Rundown AI / Articles / AI / Anthropic confronts the RSI clock

Anthropic confronts the RSI clock

PLUS: Stress test business ideas with Perplexity

Zach Mink

June 5, 2026

Good morning, AI enthusiasts. Recursive self-improvement used to sound like a far-off sci-fi phrase. Now, frontier labs say the signs are showing up internally faster than anyone expected.

Anthropic's new report shows AI already accelerating the work that builds its own successors, and lays out the risks of systems that improve themselves… While also floating the jarring scenario of a coordinated development pause across the industry.

In today’s AI rundown:

Anthropic charts path to self-improving AI
OpenAI’s memory overhaul lets ChatGPT ‘dream’
Stress test business ideas with Perplexity
Rival AI labs unite behind bioweapons risks
4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

ANTHROPIC

🔁 Anthropic charts path to self-improving AI

Image source: Anthropic

The Rundown: Anthropic just published "When AI builds itself," a report on recursive self-improving (RSI) systems — citing internal data on Claude's coding takeover and cautioning that fully self-improving AI could arrive before institutions are ready.

The details:

Anthropic noted that RSI is not here yet or even inevitable, but said that Claude is advancing AI development “faster than we thought”.
More than 80% of Anthropic's merged code was Claude-authored as of May, with engineers pushing 8x as much code per day in Q2 2026 as in 2024.
"Each new version of Claude could be built by the version before it, without human involvement," co-author Jack Clark wrote of where the trend leads.
OpenAI flagged the same loop this week in its “Democratic Governance of Frontier AI” blueprint, pointing to RSI's first sparks in today's systems.
Anthropic said it would slow or pause frontier AI if peer labs did too, and plans policy talks in the coming months to discuss research, systems, and scenarios.

Why it matters: Anthropic and OpenAI aren't alone in feeling the RSI, with labs like MiniMax saying its M2.7 model helped build itself and new startups dedicated to the self-improvement loop popping up everywhere. The unknowns of RSI are scary, but it's also hard to fathom a feasible pause scenario that hinges on global coordination.

TOGETHER WITH LIGHTFIELD

💨 The fastest-growing AI-native CRM

The Rundown: Lightfield is the CRM used by over 4,000 companies that updates itself and does the work for you. Teach it how you sell with Skills, and watch it go to work for you.

Ask it anything, like:

"Write follow-up emails to everybody I spoke with.”
"Tell me why do we keep losing to our biggest competitor."
“Find companies that look like my best customers and reach out to them.”

Try it for free at lightfield.app.

OPENAI

😴 OpenAI’s memory overhaul lets ChatGPT ‘dream’

Image source: OpenAI

The Rundown: OpenAI just introduced a new memory update within ChatGPT centered around "dreaming," a background system that turns past chats into a running, category-sorted profile of who you are for better personalization and evolving context.

The details:

ChatGPT now keeps a running written summary of each user, grouped into areas like travel, hobbies, and work, replacing the previous list of one-off facts.
Users can review memories, make corrections, add details, or ask ChatGPT not to bring up certain topics, with memory automatically updating over time.
OpenAI says factual recall rose from 41.5% to 82.8% in its evals with dreaming, while preference-following climbed from 31.4% to 71.3%.
Dreaming is rolling out to Plus and Pro users in the U.S., with Free and Go, and more countries getting the upgrade over the next several weeks.

Why it matters: Memory is one of the stickiest AI features in theory, but it’s been long overdue for an upgrade. Sam Altman frequently talks about the future of hyper-personalized AI, and dreaming may become a big part of forming that continuity and proactivity for users — and keeping them from switching to rivals in the process.

AI TRAINING

🤔 Stress test business ideas with Perplexity

The Rundown: In this guide, you will learn how to use Perplexity Deep Research to stress test any business idea. Save the prompt below once and rerun it on every idea you have to see what’s feasible to build.

Step-by-step:

Open Perplexity and switch to Deep Research mode. This works on the free plan (5 queries/day) and is basically a hidden version of Perplexity Computer
Paste this prompt with your idea in the chat, hit run, and walk away for 5 to 6 minutes. Perplexity does the research and builds the slide deck in the same run
Save the prompt somewhere you will actually use it again, like in a dedicated Perplexity space
Then, every Saturday morning, take one idea off your list and run it. You will burn through a year of half-evaluated ideas in a month

Pro tip: Build variants. A 6-slide version for a co-founder pitch, a version that compares two ideas, or a 90-day MVP plan for ideas that already cleared validation.

PRESENTED BY GITLAB

📡 GitLab Transcend goes live

The Rundown: GitLab Transcends is a live event streaming from London June 10-11, with an agenda built for developers – featuring live demos, real-world agentic AI use cases, and developer-first sessions.

With GitLab Transcends, you’ll get:

Live demos of the Duo Agent Platform
Agentic AI use cases from your peers
The Developer Show, hosted by Snerio Developer Advocate, Colleen Lake

Come see intelligent orchestration, now with context. Register free.

AI & SAFETY

🧬 Rival AI labs unite behind bioweapons risks

Image source: Open letter to Congress

The Rundown: CEOs of OpenAI, Anthropic, Google DeepMind, and Microsoft just signed an open letter pressing Congress to make synthetic-DNA sellers vet every buyer/order, warning that AI can now enable bad actors to create bioweapon designs.

The details:

Signees include Sam Altman, Dario Amodei, Mustafa Suleyman, Alexandr Wang, and Demis Hassabis, alongside DNA-synthesis industry leaders.
The letter said, “AI systems now outperform PhD-level virologists… about highly technical lab procedures in their own domains of expertise.”
The signers urge Congress to make U.S. synthetic-DNA and RNA sellers screen orders, verify buyers, and log sales to keep dangerous sequences traceable.
They also warned that "knowledge barriers which have historically prevented bad actors from obtaining biological weapons will meaningfully erode."

Why it matters: Like RSI, biological weapons concerns have been outlined for years as one of the steps up the ladder of AI improvement — and this one’s concerning enough to even unite Altman and Amodei. It’s clear that laws and regulations are going to have to adapt quickly to a changing AI-accelerated world, but the question is if they can.

QUICK HITS

🛠️ Trending AI Tools

🌌 Reve 2.0 - 4K image model with new layout-based editing
🗣️ Miso One - Open TTS model that reads tones for expressive responses
🧾 Stack - Ramp's AI accounting OS for monthly bookkeeping
🧠 Nemotron 3 Ultra - Nvidia’s open 550B reasoning model for agents

📰 Everything else in AI today

The U.S. and Japan announced a $1B AI research partnership, becoming the first country to join the U.S. Genesis Mission’s push to double science output using AI.

Nvidia released Nemotron 3 Ultra, a fully open 550B reasoning model that runs 5x faster and up to 30% cheaper for agents, with performance similar to top open rivals.

Canadian Prime Minister Mark Carney introduced AI for All, a new five-year national strategy surrounding the tech that targets $200B in growth and 250,000 AI jobs.

Gopuff launched Go, an AI shopping assistant built with SpaceXAI that predicts users’ carts using Grok, real-time signals from X, and order data.

CloudFlare co-founder Matthew Prince revealed that bot traffic on the internet has now surpassed humans, with the milestone coming a year ahead of his expectations.

COMMUNITY

🤝 Community AI workflows

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader Anonoymous:

"I have 2 small kids at toddler age, and at times that can be very, very challenging. I have a scheduled task to run a pre-set prompt a couple of times a week to provide tips and tricks that can help me manage the day-to-day.

Sometimes it suggests an activity that I can try on the weekend, other times it suggests how to handle poor behaviour, time-outs, or how to handle bedtime routine. In sum, it helps expand the options available to me on how to deal with lots of different sides of parenting, helping me become a better father."

How do you use AI? Tell us here.