In partnership with
Welcome to OHMO AI. The government that banned Anthropic's strongest AI just handed it back, to a hundred companies it won't name. OpenAI's newest model is stuck in the same waiting room. And Anthropic says someone copied Claude using nothing but questions. Who gets the most powerful AI is now a decision made in Washington.
In today’s newsletter:
The US lets Mythos 5 back out to a vetted set of organizations
Anthropic accuses Alibaba of copying Claude through millions of questions
OpenAI previews GPT-5.6 but holds the public launch at the government's request
Anthropic's H-1B filings reveal base salaries up to 1.38 million dollars
LATEST IN AI
ANTHROPIC
Mythos 5 Comes Back to a Chosen Few
Image source: Anthropic
OHMO AI: The US government has partially lifted its ban on Anthropic's most powerful AI, telling the company on Friday that Mythos 5 can be redeployed to organizations that operate and defend critical infrastructure. More than 100 companies and agencies, including many Fortune 500 names, regain access, though the model stays closed to the public.
The details:
The reversal came in a letter from Commerce Secretary Howard Lutnick, who wrote that appropriate safeguards were now in place for certain trusted partners.
The original June 12 order had pulled both Mythos 5 and Fable 5 offline, even barring non-American staff at Anthropic and partner organizations from access. That staffing restriction is now lifted for the approved group.
Anthropic says it's still working with the government to widen Mythos 5 access and to bring Fable 5, its more guarded public model, back to general use. No timeline was given.
Why it matters: Frontier AI is no longer something a company simply ships. A model this capable now reaches users only through a government list, and nobody outside the room knows how that list is drawn. OpenAI's Sam Altman said he doesn't mind safety testing but objects to the government picking who gets the tools. That argument is just getting started.
TOGETHER WITH WISPR FLOW
The best prompt engineers aren't typing. They're talking.
Power users figured this out early: speaking a prompt gives you 10x more context in half the time. You include the edge cases, the examples, the tone you want — because talking is fast enough that you don't skip them.
Wispr Flow captures everything you say and turns it into clean, structured text for any AI tool. Speak messy. Get polished input. Paste into ChatGPT, Claude, Cursor, or wherever you work.
89% of messages sent with zero edits. 4x faster than typing. Works system-wide on Mac, Windows, and iPhone.
ANTHROPIC
Alibaba Allegedly Talked Claude Into Talking
Image source: OHMO AI
OHMO AI: Anthropic has told US lawmakers that operators linked to Alibaba and its Qwen AI lab ran the largest known effort to extract Claude's capabilities, using roughly 25,000 fake accounts to generate more than 28.8 million interactions. The goal, Anthropic claims, was to copy Claude's reasoning to improve Alibaba's own models. Alibaba denies it, and the allegations haven't been independently verified.
The details:
The campaign ran from April 22 to June 5, according to a letter reviewed by Reuters, and targeted Claude's software engineering, agentic reasoning, and complex task abilities rather than ordinary chat.
The technique is called distillation, where one model is trained on the outputs of a stronger one. It's common when a company does it to its own model, but Anthropic argues this was unauthorized extraction of its work.
Anthropic has made similar accusations against DeepSeek, Moonshot AI, and MiniMax earlier this year, and asked lawmakers to act on the problem quickly.
Why it matters: If you can rebuild a rival's AI just by asking it enough clever questions, the billions spent training a frontier model start to look fragile. The next fight in AI may not be about who builds the smartest model, but about stopping everyone else from quietly learning how it works.
OPENAI
OpenAI's Newest Model Hits a Gate
Image source: OpenAI
OHMO AI: OpenAI is previewing GPT-5.6, its strongest model yet, but holding back the full public launch at the US government's request. The flagship, called Sol, is going to a small group of trusted partners first, with general availability promised in the coming weeks.
The details:
The preview launches alongside two other tiers, Terra for everyday work and Luna for low cost, with the model available through the API and Codex to select partners at the start.
OpenAI says it shared the models' capabilities with the government ahead of launch, and started the limited release at the administration's request. The company added it doesn't think this kind of government access process should become the default.
On a cyber benchmark called ExploitBench, OpenAI reports GPT-5.6 Sol is competitive with its earlier Mythos Preview comparison while using far fewer tokens. That's a vendor benchmark, not an independent test.
Why it matters: Two of the biggest labs are now releasing their best models on Washington's schedule, not their own. The same week the government reopened Anthropic's Mythos, it slowed OpenAI down. The pattern is clear, and the people who actually use these tools are the last to know the rules.
PRESENTED BY PODIUM
Their first after-hours call was a $20,000 job.
Air Texas was paying $2,000 a month for an answering service that couldn't close jobs.
Their first after-hours call with Podium’s AI Employee booked a $20,000 job.
Now no call goes unanswered after 5PM.
ANTHROPIC
Inside Anthropic's Eye-Watering Pay
Image source: OHMO AI
OHMO AI: Federal visa filings show Anthropic is paying top technical staff base salaries above 1.3 million dollars, according to a Business Insider analysis of the company's H-1B sponsorships. Even finance and accounting roles can reach 200,000 dollars in base pay.
The details:
Members of technical staff showed listed base salaries ranging from about 134,000 dollars to 1.38 million dollars, though the filings don't reveal what each role actually does.
These figures are base salary only. They exclude bonuses and stock options, which at a company valued near a trillion dollars can make up a large share of total pay.
Anthropic has been poaching talent from Google and OpenAI, and keeps a high retention rate that CEO Dario Amodei has put down to belief in the mission and the upside of the equity.
Why it matters: The AI talent war has a price tag, and it's printed in government paperwork. When a single engineer's base pay tops a million before equity, it shows how few people can actually build these systems and how hard the labs will fight to keep them.
AI TOOLS
Useful AI Tools
Base44: No-code AI app builder that turns your words into fully functional apps in minutes.*
Chatbase: Build and deploy custom AI support agents for your business, no coding needed.*
ElevenLabs: Generate realistic, studio-quality voiceovers from text in any language or voice.*
Flora: Node-based creative canvas built for professional designers, agencies, and creative teams.*
Kick: Automates bookkeeping and tax preparation for business owners using AI.*
Magica: All-in-one AI platform with 5000+ tools for text, images, videos, audio, and lots more.*
WisprFlow: Voice-to-text AI that turns your speech into polished writing, so you can just talk instead of type.*
*Sponsored Listing
COMMUNITY
Featured Reader
Every newsletter, we feature how a reader is using AI to work smarter, save time, or make life easier.
Today's reader is Bianca F. from Lisbon, Portugal:
"I run a small ceramics studio and I teach evening classes. The making I love. The admin nearly closed me down. Class bookings came through Instagram DMs, email, and a notebook by the till, and I'd double-book people or forget who'd paid. I sat with AI and described the mess, and it helped me build a simple booking sheet that takes the day's messages and lays out who's coming, who's owing, and which kiln slot they need. I check it with my coffee every morning. Last term I stopped losing spots, and I actually filled two classes I used to leave half empty. A potter friend asked who set it up for me. Nobody did."
Want to be in the next newsletter? Tell us how you use AI here.
AI Meme
Go Deeper
AI Tools: A curated list of the AI tools worth knowing.
Instagram: The AI page they pray you’ll never find.
Facebook: For the ones who never left Facebook.
