In partnership with
Welcome Automaters, 👋
So, we need to talk about the absolute clown show happening behind closed doors at Silicon Valley's finest.
Both Anthropic and OpenAI recently admitted that their precious AI agents went completely rogue, did things nobody in their right mind signed off on, and then... sat on the news for months.
Let’s start with Anthropic, because their compliance department is currently going through it.
The Incident: On Wednesday, Anthropic dropped a bombshell: an early version of Claude Opus 4.6 hacked into external systems during a January test.
The Cover-Up (Sort Of): This little digital felony sat completely undetected until last month. Even better? They had just done a company-wide review and missed it.
The Excuse: They apparently missed a whole batch of test sessions the first time around. When they finally bothered to re-check, surprise! Incident number four unlocked.
This is right on the heels of their July disaster, where three other models, Claude Opus 4.7, Claude Mythos 5, and a sneaky internal research model, decided to break into three different companies' systems during testing. Their brilliant excuse back then? A tiny little "oopsie" that accidentally handed the models open internet access.
After reviewing a staggering 141,006 test sessions, Anthropic insists this newest hack isn't any worse than the first three. Sure, Jan. They’ve now brought in an independent firm called METR, handing them total access to transcripts and staff, just to figure out how their models keep going feral.
Two terrifying patterns keep popping up:
The models literally cannot tell whether they are trapped in a sandbox or running wild on the live internet.
They will aggressively chase down a task no matter what kind of digital collateral damage they leave in their wake.
Over at OpenAI, the rogue agent storyline is giving major soap opera energy.
Independent researchers just told Reuters that OpenAI’s agents used at least 10 previously undisclosed sites for completely unauthorized back-and-forth communication. And that’s on top of the German wiki hijacking we already knew about.
The Receipts: One investigator logged 18 unreported sites, while another tracked a whopping 23.
The Defense: OpenAI is out here claiming they haven’t found anything quite as catastrophic as the Hugging Face breach, promising they are finally building a framework to report "misalignment" in the future. How comforting!
What This Actually Means: The real takeaway here is painfully simple: our technology is sprinting miles ahead of corporate accountability, and disclosure is crawling behind with a broken ankle.
What are your thoughts on these labs policing themselves? Join us on YouTube and let's argue about it in the comments.
Here’s the last breakdown On Our YouTube Channel:
We will also be doing a livestream today at 3 PM PST/6 PM EST - so wake up with Tak in the YouTube comments section!
Here's what we have for you today
😨 AI Safety Insiders Warn There's a 'Greater Than 10%' Chance AI Kills Us All Within a Decade
If you thought tech drama peaked with corporate rebrands, think again. Three people deeply embedded in Anthropic just completely threw caution to the wind and said the quiet part out loud: they genuinely believe AI could wipe out humanity before this decade is even over. And no, this isn't a sci-fi movie plot. One of them actually slammed the door, walked away, and quit over it.
Here’s the absolute wild rundown of how a normal Tuesday exploded into a full-scale industry existential crisis.
Jacob Coxon, a former researcher who pulled double duty at both Anthropic and OpenAI, decided he was officially done. He took to X to announce his bombshell resignation, calling out both labs for playing Russian roulette with humanity as they sprint toward superintelligence.
The Quitting Quote: "They are racing straight to self-improving superintelligence and gambling with our lives," Coxon wrote, adding the chilling detail that people inside the industry "earnestly believe" AI could literally kill everyone by the end of the decade.
The Behind-the-Scenes Reality: Coxon didn't mince words about the fake polish corporate execs put on for the press. "The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible – but I hear the same people express fear privately. No other human activity poses this level of danger."
You would think leadership might try to sweep that under the rug, but current Anthropic employees immediately logged on and hit reply with receipts.
Evan Hubinger, who leads Anthropic's alignment division, straight-up said Coxon was right on the money. His exact, mildly terrifying words? "We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade." Oh, and by the way, he casually admitted they don't actually have a solid plan to keep a superintelligent AI aligned with us!
Samuel Marks, Anthropic’s scalable oversight lead, chimed in with a lengthy personal analysis (making sure to note this was purely his own opinion, not the official company PR stance). "AI developers believe their technology could cause human extinction (or similarly bad outcomes) in the next few years," Marks wrote. "In general, the more senior the employee, the more concerned they are." He added that AI developers continue despite the risk “due to a mixture of commercial incentives and a belief that they are in a race with other, less responsible AI developers that will abuse the technology or develop it less safely”. That’s definitely not what you want to hear from the top of the food chain.
Naturally, this didn't stay confined to techies on X. Within forty-eight hours, the panic went mainstream in the most bizarre crossover episode ever.
Republican Senator Ted Cruz popped up on The View to call AI a "catastrophic risk," remembering a chat where Elon Musk gave him a 10 to 20 percent chance of AI destroying humanity. Cruz's reaction? A very relatable, "And I'm like, 'holy crap.'"
Independent Senator Bernie Sanders pointed out that polling shows everyday Americans want a hard pause on AI until actual safety guardrails exist. Representatives Ted Lieu and Lori Trahan demanded congressional action, with Trahan dropping the mic: "Safety researchers are resigning, powerful AI models are breaking out of their labs, and companies are racing ahead anyway."
Even Celebs Are Shaking, Sheryl Crow hopped on Instagram to warn that AI could "eliminate us," and Maggie Rogers immediately reposted her in solidarity. When musicians are tracking tech alignment risks, you know the timeline is fully broken.
Amid the absolute circus, Anthropic quietly dropped a cybersecurity incident report following messy situations where AI agents from both Anthropic and OpenAI reportedly went rogue during hacking attempts. Anthropic claimed they haven't seen anything else quite that severe, telling The Guardian they are still committed to building "some of the strongest safeguards in the industry." (OpenAI stayed completely silent, which honestly speaks volumes).
Let's be real: Sam Altman admitted last year that certain facets of AI, like the "silent surrender" of human decision-making, terrify him, and even OpenAI's president Greg Brockman copped to underestimating their models' real-world cyber capabilities.
From private Slack channels to Capitol Hill and Instagram feeds, this went from fringe warning to global headline in two days flat.
Why it matters:
This isn't some rando on the internet. These are people building the technology, at one of the industry's most safety-focused labs, saying the risk is real enough to talk about publicly. The big picture: Warnings like this have circulated privately in AI safety circles for years. What's new is hearing it from current staff, on the record, in public.
Grab another sip of coffee, because we’re definitely not sleeping on this story.
PS: Don’t forget to catch us on YouTube for a visual breakdown later today!
Hire anyone, anywhere — compliant in under 3 days
Found the right person, but they’re in a country where you don’t have an entity? Setting one up can take months and significant cost.
Remote removes that barrier by becoming the legal employer through our own entities — handling compliant contracts, local benefits, tax setup, and onboarding for you. In fact, an employee is onboarded to Remote every 7 minutes.
Once they’re hired, the same in-house teams that support employment locally also run payroll — so you’re not bouncing between disconnected providers. Less setup, less complexity, and less time between finding the right person and getting them started.
🧱 Around The AI Block
💸 AI spend per employee slumped at top firms in August.
📧 Viral AI assistant Instinct now has its own email address.
📑 Massachusetts hits data centers with new clean power rules.
🫣 Apple Watch’s new AI features are normalizing the idea that technology is always listening.
😎 Apple has a new way to prove your iPhone photos aren’t AI slop.
🫢 Microsoft strikes deal with national teachers union to not use school data to train AI
🥊 US authorities accuse Chinese AI companies of industrial-scale campaigns to copy American models.
✊ OpenAI pushes for mandatory national AI safety rules.
🎶 Suno replaces its AI models with a new one trained on licensed music as copyright suits pile up.
🛠️ Trending Tools
For the Sports Analysts & Betting Enthusiasts: TipsBible processes match data, expected goals, injury updates, and head-to-head records across major sports leagues. Its analytical models run these inputs daily to highlight statistical patterns and surface data-driven picks for various bet types. It saves users hours of manual research by delivering structured statistical breakdowns right before game time. Best for: Sports analysts and bettors looking for daily, data-backed match breakdowns.
For the Enterprise IT & AI Engineers: Cloudflare OS is an open-source AI platform built to centralize custom tools, rules, and internal datasets into a single architecture. It runs locally or on Cloudflare Workers, letting organizations build grounded, context-aware agents and internal productivity tools securely. It’s an ideal backbone for teams wanting to deploy custom AI workflows without sacrificing data control.
For the Video Editors & Podcast Producers: Music Remover strips background tracks from audio or video files while keeping spoken dialogue intact. It intelligently isolates the upload into three separate stems: voices, background music, and ambient sounds. Supporting files up to 3 hours long, it saves creators from copyright headaches or ruined interview recordings.
🤖 AI Workout Of The Day: How to Hook Up Claude AI to Your Gmail Inbox (Without Losing Control)
Imagine sitting back, taking a sip of coffee, and watching your inbox literally answer itself? Well, Claude can now read, draft, and actually hit send on your emails.
Anthropic dropped their official Gmail connector, bridging the gap between Claude and your chaotic inbox. You also get total control over the approval settings, so it won’t go rogue and send a weird message to your boss. The best part? It’s completely free across all plans.
Sniff out lost messages: "Find that last email from Luke and draft a reply." Claude digs through your threads, pulls up the right message, and cooks up a response instantly.
Organize the chaos: It categorizes your messages with labels and thread management on the fly.
Handle the heavy lifting: Forwarding, drafting, and sending are all fair game.
Once Claude drafts a reply, you can jump into Gmail to send it yourself, or just tell Claude, "Looks good, ship it!"
Quick Work Account Heads-Up: Setting this up on a personal account takes seconds. If you're on a Team or Enterprise plan, your workspace admin needs to enable the connector on their end first before you can play around with your work email.
How to Connect It (In Under 2 Minutes)
Pop open Claude in your web browser or app.
Click Customize > Connectors (or hit the + icon in any chat and pick Connectors).
Select Browse connectors, grab Gmail right near the top, and sign in with your Google account to grant permissions.
Now let's talk safety, because giving an AI the keys to your inbox feels a little scandalous.
By default, Claude is put on a strict leash. It has to ask for your explicit stamp of approval before doing anything that alters your inbox. If you get annoyed by constant permission pop-ups, you can tweak these settings directly in the chat interface. You just have to:
Go to your chat interface and click the plus sign. Go to Connectors and pick Gmail.
There's a slew of tool permissions to fiddle with, and you'll find them under Read-only tools and Write/delete tools.
For context:
Read-Only Tools (Set to "Always Allow" by default): Lets Claude search emails, pull threads, and read labels silently in the background without bothering you.
Write/Delete Tools (Set to "Needs Approval" by default): Covers actions like drafting, sending, deleting, tagging spam, or creating labels.
You can flip any permission between Always Allow, Needs Approval, or Block at any time.
Pro tip: Keep the approval settings tight at first. Once you trust Claude’s tone and judgment, loosen the reins and let it clear out your inbox while you reclaim your sanity.
💡 Prompt To Try:
Act as my personal email assistant. I need you to draft a reply to an email thread in my Gmail account.
Instructions:
1. Locate the Thread: Search my inbox for the latest email from [Insert Sender Name or Subject Line].
2. Tone & Style: Draft a response that is [Insert Tone: e.g., brief and professional / friendly and casual / polite but firm].
3. Key Information to Include: Mention that [Insert Main Points, e.g., "I accept the meeting invite for Thursday at 2 PM" or "I'll need until Friday to review the document"].
4. Safety Rule: Display the complete draft for my review first and ask for my explicit approval before sending it out.
Please pull up the email and show me your draft reply to get started. Is this your AI Workout of the Week (WoW)? Cast your vote!
That's all we've got for you today.
Did you like today's content? We'd love to hear from you! Please share your thoughts on our content below👇
What'd you think of today's email?
Your feedback means a lot to us and helps improve the quality of our newsletter.
