- Everyday AI
- Posts
- Ep 801: Fable 5 Drama Updates: The latest Between Trump vs. Anthropic and How it Impacts You
Ep 801: Fable 5 Drama Updates: The latest Between Trump vs. Anthropic and How it Impacts You
Google Gemini co-lead Noam Shazeer leaves Google for OpenAI, Anthropic says access to Fable 5 could return in days, OpenAI is bringing scheduled tasks back to ChatGPT and more.
Everyday AI
June 18, 2026
š Subscribe Here | š£ Hire Us To Speak | š¤ Partner with Us | š¤ Grow with GenAI
Sup yāall š
On todayās show, we gave you the tl;dr version of the Anthropic vs. U.S. Government fight over Fable 5.
Whatās your take?
What's your take on the Anthropic vs. U.S. Gov fight on Fable 5 and Mythos 5 pulled access?š³ļø Vote to see LIVE results š³ļø |
āļø
Jordan
P.S. - reach out on LinkedIn and tell me youāre from the newsletter!
Outsmart The Future
Today in Everyday AI
8 minute read
š Daily Podcast Episode: Anthropic is still battling U.S. restrictions on Claude Fable 5, and the fallout is raising bigger questions about who controls the most powerful AI models. Give todayās show a watch/read/listen.
šµļøāāļø Fresh Finds: Claude Design is getting major upgrades for teams, GPT-5.4 helped improve a real-world drug development process, and OpenAI is preparing a new ChatGPT plan for scientific research. And more. Read on for Fresh Finds.
š Byte Sized Daily AI News: Google Gemini co-lead Noam Shazeer leaves Google for OpenAI, Anthropic says access to Fable 5 could return in days, OpenAI is bringing scheduled tasks back to ChatGPT and more. Read on for Byte Sized News.
šŖ Leverage AI: Anthropic's Fable 5 shutdown showed how quickly access to a frontier AI model can disappear, even after a major launch. Keep reading for that!
ā©ļø Donāt miss out: Miss our last newsletter? We covered: Report: OpenAI readying new groundbreaking voice model, Fable 5 fight goes global, Cursor teases big new AI model and more. Check it here!
Ep 801: Fable 5 Drama Updates: The latest Between Trump vs. Anthropic and How it Impacts You
The Fable 5 drama continues.... but what does it really mean? š¤
Anthropic released the most powerful AI model in the world, only for it to be pulled worldwide days later over bio and cyber concerns.
Sound like a tough break, right?
But, was the writing on the wall all along? Was Anthropic asking for this type of regulation? And what does the future of heavy-handed government regulation mean for your company and its AI strategy?
Also on the pod today:
ā¢ Fable V: Pulled after just 3 days online šØ
ā¢ Andy Jassy: triggered the government shutdown šØ
ā¢ David Sacks vs. Anthropic: he said he said š¤Ø
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Hereās our favorite AI finds from across the web:
New AI Tool Spotlight ā Upstream organizes your inbox, cuts the noise, drafts replies and follow-ups in your voice so you feel lighter and focused, Jesse crawls the live internet to surface leads who are ready to buy right now, Adapt is the company brain: one shared context layer for every employee and every agent that answers questions and does work
Claude Design Upgrades ā Claude Design now lets you build with your own design system, sync seamlessly with code, and export to your favorite tools.
Character.ai Upgrades ā Character.ai just dropped creator tools like guides, insights, and leaderboards to help you make better Characters and grow your audience.
Adobe Sued by Investors ā Adobeās execs claimed their AI didnāt use copyrighted material, but investors say the truth cost billions and triggered lawsuits.
ByteDance Seedance 2.5 Tease ā ByteDance's Seedance 2.5 is dropping in early July with longer video support, and 2.0 is about to get 4K upgrades.
GPT-5.4 and Chemistry ā GPT-5.4 helped chemists boost a tough drug-making reaction with real lab results, not just theory.
Americans and AI ā Half of U.S. adults now use AI chatbots, but most think AI is moving too fast and will make their data less secure.
Amazon NVIDIA and AMD ā Amazon, Nvidia, and AMD just poured $310 million into Odyssey ML to build AI that understands the real world, not just words.
OpenAI LifeSciBench ā LifeSciBench puts AI to the test with real-world life science research challenges, and shows that while models are getting better at scientific reasoning, they still struggle with complex experiments and artifact-heavy tasks.
xAI and Amazon ā xAIās Grok 4.3 just landed on Amazon Bedrock, giving AWS developers access to top tool-calling and low hallucination rates.
New Google Home ā Googleās new Home Speaker brings natural conversations and smarter controls to your home with Gemini.
ChatGPT for Science ā OpenAI is gearing up to launch a "ChatGPT for Science" plan aimed at research institutions. No details yet, but it's set to rival Google's and Anthropic's science offerings.
MolmoMotion Model ā MolmoMotion predicts 3D object motion from video and instructions, outperforming other models in robotics and video generation.
1. Googleās Gemini co-lead Noam Shazeer is leaving for OpenAI š±
Reuters reports that Noam Shazeer, a Google VP of engineering and co-lead of its Gemini models, said Wednesday he is leaving to join IPO-bound OpenAI.
The move is striking because Google reportedly spent $2.7 billion less than two years ago to bring Shazeer and his Character.AI team back, underscoring how fiercely top AI labs are fighting for talent.
2. Anthropic says blocked Mythos and Fable 5 access could return within days š§
Anthropic told reporters in Seoul on Wednesday that access to its restricted frontier models could be restored āin the coming days,ā after a U.S. security directive led the company to block them globally.
The move has put Korea in the spotlight because a Korean telecom firmās reported access to Claude Mythos helped trigger Washingtonās scrutiny over potential China-linked risks.
3. Midjourney announces ultrasound body scanner and 2027 San Francisco spa rollout š©»
Midjourney, best known for image generation, says it is building a water-based medical imaging scanner designed to create fast 3D body maps using ultrasonic sensors, with its first āresearch spaā planned for San Francisco in 2027.
The pitch is ambitious: make body scanning feel more like a quick spa visit than a hospital appointment, starting with body composition maps while pursuing FDA clearance for broader medical uses.
4. OpenAI brings scheduled tasks to ChatGPT today šļø
OpenAI is rolling out scheduled tasks in ChatGPT today for Plus, Pro, Business, and Enterprise users, giving the app a more practical way to handle reminders, recurring work, and watch-for-updates requests.
The feature adds a new Scheduled page where users can see what is active, when it runs next, and quickly pause, edit, resume, or delete tasks.
5. Anthropic and DeepMind CEOs push U.S.-led AI rules at G7 summit š«ø
According to CNBC, Anthropicās Dario Amodei and Google DeepMindās Demis Hassabis urged G7 leaders on Wednesday to back a U.S.-led coalition for setting global AI rules and safety standards.
The push comes just days after the U.S. imposed export controls on Anthropicās newest models, reflecting growing concern that advanced AI tools could be misused in cyber, biosecurity and intelligence operations.
6. Elon Musk predicts Chinese āFable-Classā Model by Q1 2027 šØš³
A live X exchange today put a date on the global AI race: Elon Musk expects a Chinese āFable-classā model by Q1 2027, and Z.ai publicly accepted the challenge.
Teortaxes framed the timeline as aggressive, noting Chinaās leading labs are only expected to start building much larger clusters this fall and have not yet trained a 1T-parameter model.
Your business momentum can get hijacked by someone elseās AI chess match.
Thatās the real Anthropic lesson.
Fable 5 and Mythos 5 didnāt just get pulled after a fight over cyber risk and foreign access. They showed how fast your AI-driven progress can get trapped between vendor hype, regulator pressure, cloud partner warnings, geopolitical risk, and pre-IPO gamesmanship.
Youāre trying to move faster. The labs are trying to win markets, shape policy, impress investors, slow competitors, and still sell you the tools powering your next workflow overhaul.
Awkward little setup.
Thatās what we tackled today on Everyday AI: how Anthropicās own safety messaging became ammunition, why frontier-model access now belongs in business continuity planning, and how leaders can keep AI progress from becoming collateral damage in someone elseās valuation play.
1. Audit safety claims before approving critical workflows š„
Anthropic spent weeks making Mythos 5 sound almost too powerful to release. Then came Claude Opus 4.8, IPO paperwork, an industry slowdown call, Fable 5, and the global shutdown.
That sequence should make execs slow down for about 12 seconds.
When a vendor markets a model like it belongs behind glass, regulators donāt need to invent the threat narrative. They can just recycle the sales pitch with scarier letterhead.
The business risk is bigger than bad PR. Safety blogs, model cards, policy posts, and executive interviews can turn into procurement friction, legal exposure, access restrictions, and board-level questions about why your team built important work on a model wrapped in warning tape.
Try This
Before approving a frontier model for real workflows, create a vendor narrative file. Drop in the launch post, model card, safety post, access terms, and enterprise pitch, then highlight every line that suggests cyber risk, weapons framing, restricted access, foreign access limits, or ātoo powerfulā energy.
Then ask: if regulators believed every word, would our rollout still look responsible?
2. Build fallback lanes before frontier access vanishes ā”
Fable 5 was public for about three days.
Thatās barely enough time for your AI lead to hype it in Slack, much less rebuild a customer-support flow, sales workflow, coding process, or internal ops system around it.
Enterprise AI access can disappear through government action, not just vendor failure. A model can still work perfectly while your business loses the right to use it.
That changes the job.
Leaders canāt just ask which model is best. They need to know which model is stable, which model has a tested fallback, which workflows can safely use frontier access, and which ones should stay on boring-but-reliable infrastructure.
Try This
Run a āmodel disappears tomorrowā drill on your five highest-value AI workflows. For each, name the primary model, fallback model, data path, owner, compliance issue, manual process, rollback trigger, and testing cadence.
Then test the backup with real work. If the fallback hasnāt survived a live task, itās not a fallback.
3. Separate business progress from frontier politics š
Anthropic may have been safety-first, strategically theatrical, or just weirdly comfortable lighting its own launch on fire. Doesnāt matter enough for your Monday morning.
Your job isnāt to defend a labās storyline.
Your job is to protect business progress while Anthropic, OpenAI, Google DeepMind, Amazon, regulators, governments, investors, and national security teams all tug on the same frontier stack.
That means ābest modelā is too shallow. The better executive question is which model can drive measurable progress, survive access shocks, pass governance, and swap out cleanly when the AI circus finds a new tent to burn down.
Try This
Create three model lanes this week: production, experimental, and restricted. Production models get fallback plans, access reviews, audit trails, export checks, and clear owners; experimental models get sandboxed use cases and short review windows; restricted models need legal, security, and executive signoff before anyone builds around them.
Use frontier models to push progress.
Donāt let them own it.
Reply