Good morning {{first_name| Artificial intelligence lovers}}. The government has spent months escalating its battle with Anthropics. Then Mythos showed up with cybernetic abilities powerful enough to make the dispute seem less simple.
Now the White House is trying to thread a strange needle: keeping the model close to national security, limiting who can use it, and avoiding appearing to completely back away from the Pentagon’s hard line, all at the same time.
-
The White House’s humanitarian position becomes complicated
-
Gemini comes in cars powered by Google engines
-
Stress test business ideas with bewilderment
-
OpenAI finds the source of ChatGPT’s daemon obsession
-
4 new tools for AI, community workflows, and more
Anthropic vs. The White House
Image source: Images 2.0 / Rundown
Rundown: The white house is Push back About Anthropic’s plan to more than double private sector access to Mythos AI for computing concerns for its own use, just as for national security note He prepares to address parts of the Pentagon dispute.
-
The requested humanitarian access expanded from about 50 companies to nearly 120 companies, as U.S. officials cited computing pressures that could impact government use.
-
The White House memo on AI will reportedly push agencies to adopt multi-vendor AI and address some of the Anthropic concerns that led to the initial move runner.
-
Axios I mentioned The government action “will allow agencies to circumvent supply chain risk classification,” despite the current legal battle.
-
GPT 5.5 has been reached similar Cyber Capabilities to the Mythos, with former AI czar David Sachs Saying All border models will level up within 6 months.
Why it matters: The White House is changing its tune toward Anthropists, apparently in part because it wants greater access to powerful myths. But with a second. Of the war Pete Hegseth said Thursday that while Anthropics is “run by an ideological lunatic,” there is some internal division between a desire to bury the hatchet versus continuing the fight.
TOGETHER WITH GOOGLE FOR STARTUPS
Rundown: Synthesia’s CEO reveals how the disappearing costs between text and video will soon replace standard slide decks with custom, real-time multimedia for enterprise operations.
In this new report, you’ll hear views on:
-
The vanishing cost gap between text and video
-
Dynamic multimedia replaces standard slide decks
-
Design timely content creation pipelines
-
Simplifying knowledge exchange processes between companies
Rundown: Google is beginning Gemini upgrades Google’s built-in vehicles, replacing Assistant with a more conversational system that handles navigation, messaging, music, vehicle questions, and hands-free controls across compatible cars.
-
Drivers can request changes to vehicle settings such as temperature and radio control and pull up from Google Maps for personalized updates or route planning.
-
Beta Gemini Live mode supports conversations for learning and brainstorming, with the ability to integrate Gmail, Calendar and Home later.
-
Gemini can also pull vehicle-specific answers from manufacturer manuals for vehicle assistance and battery status or charging stations for electric vehicles.
-
This rollout to compatible vehicles in the US comes first, as GM also announced the feature for approximately 4 million of its vehicles from the 2022 model year onwards.
Why it matters: One day, AI integrations in cars will be as common as radio (and eventually, the systems will all drive cars, too) – but for now, we’re still in the rollout phase. These initial features are fairly basic, but they’re a step on the way toward AI-era “smart car” systems that provide a serious upgrade to intelligence.
Artificial intelligence training
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 replay it over each idea you have to see what can be built.
-
It opens Confusion And go to deep search mode. This works on the free plan (5 queries/day) and is basically a hidden copy of Perplexity Computer
-
sticky This claim With your idea in the chat, press play, and step away for 5 to 6 minutes. Perplexity researches and creates the chipset in the same run
-
Save the claim somewhere you’ll use it again, such as in a dedicated Perplexity space
-
Then, every Saturday morning, cross one idea off your list and run with it. You will burn through a year of half-evaluated ideas in a month
Pro Tip: Construct variables. A 6-slide version to show a co-founder’s pitch, a version comparing two ideas, or a 90-day MVP plan for ideas that have already been validated.
Brought to you by FUEL IX by TELUS DIGITAL
Rundown: Uncharted: AI Safety and Security Summit unites world leaders to bridge the gap between rapid innovation and production-level governance. Join Fuel iX on May 5 to cut through the hype and explore the technical frameworks and proprietary research required to secure the next generation of enterprise GenAI.
-
Close the safety gap without slowing down technical innovation
-
Learn practical, proven strategies to protect users and secure AI applications
-
Get exclusive insights from the benchmark safety tests of leading MBAs
Artificial intelligence research
Rundown: OpenAI only tracking ChatGPT’s habit of mixing up her responses with orcs, gremlins, and various fantasy creatures goes back to one bonus reference in her “obsessive” personality, which ended up becoming typical behavior throughout the issues.
-
Following the launch of ChatGPT-5.1 in November, mentions of the word “goblin” jumped 175% in user conversations, with “gremlin” up 52% and other creatures seeing similar spikes.
-
When OpenAI mapped the use of creatures across characters, the Nerdy preset lit up, pushing two-thirds of all mentions to ghouls from just 2.5% of traffic.
-
Even users who skipped Nerdy got sprites, with fine-tuning loops to recycle the creatures’ preferred output back to ChatGPT’s default.
-
OpenAI retired Nerdy in March and shipped GPT-5.5 with a Codex claim that specifically banned orcs, gremlins, ogres, trolls, raccoons, and pigeons.
Why it matters: The ChatGPT daemon mode is a fun little feature for your Friday night, and another example of how weird LLMs really are. A bonus in single character mode spawned a pattern of creature preferences that spread across chats around the world. Just like Anthropists Golden Gate Claudewe may need a standalone GoblinGPT.
-
💳 connection – Stripe wallet for AI customers with human approval on every purchase
-
🔐 Claude Security – Enterprise tool for scanning and patching vulnerabilities
-
🎨 Imagine the agent – Proxy canvas for xAI in Grok for image and video creation
-
🤖 Cloud computer – Manus cloud for permanent agents, scrapers and more
dead open Its advertising platform accesses third-party AI tools via a new MCP server, allowing advertisers to manage campaigns through Claude, Cursor, or any connected agent.
OpenAI Announce It has already surpassed Stargate’s 2029 goal of securing 10 gigawatts of compute, with 3 gigawatts added in the past three months.
Elon Musk open During cross-examination in its trial against OpenAI, xAI used distillation techniques to train OpenAI models.
Anthropic Fired Public beta release of Claude Security, a system that leverages Opus 4.7 to scan code bases for vulnerabilities and help organizations build patches.
Indicator Released Security Review, which also deploys independent agents to check vulnerabilities and conduct scheduled code base scans with the results posted on Slack.
In each newsletter we show how the reader is using AI to work smarter, save time or make life easier.
Today’s workflow comes from the reader unknown:
“Like millions around the world, I am a recovering addict, and more specifically, an alcoholic. There is a lot of literature and studies available. Some of the primary writings date back to the 1930s and have been preserved in their original form to preserve their meaning.”
The volume of information is staggering and the usage and nuance is essential to keeping the message personal. To help with my own journey, I created a notebook in NotebookLM that references several books produced by Alcoholics Anonymous, clinical research studies, and the works of independent authors.
This allows for daily motivational messages, key topics for a particular recovery step, or clarification of crucial material with pros and cons. Turning some of these topics into an audio discussion in NotebookLM is a great way to learn about different viewpoints and really see the differences and similarities. “It is also a true vehicle for speaking notes, retrieval workbooks, and deep research.”
How do you use artificial intelligence? Tell us here.
That’s all for today!
Before you go, we’d love to know what you thought of today’s newsletter to help us improve The Rundown experience for you.
Rowan, Joey, Zack, Shubham and Jennifer – the humans behind The Rundown
