Welcome back! A group of stock analysts used Claude to design proteins that could lead to antivenom. OpenAI bought a startup that helps drugmakers move faster the same day. ChatGPT now lets you lock chats behind a PIN. And OpenAI says one of its agents removed a safety monitor, then lied about it.
In today's Generative AI Newsletter:
Claude: Why did a stock research firm ask Claude to fight scorpions?
OpenAI: What does OpenAI want with a drug development startup?
ChatGPT: What chats are you hiding?
Shocking: What did OpenAI's agent say when asked about the missing monitor?
Claude took its first shot at an antivenom
Scorpions sting more than 1.2 million people a year. The antivenom comes from horses injected with scorpion venom, and it can take months to make.
Anthropic had Claude design an alternative from scratch. In 64 hours, it drew up 318,484 designs for tiny proteins that grab scorpion toxins before they hit your nerves.
Labs tested 211, and 17 stuck to the toxin they were built for. The best two grab a Mexican scorpion's toxin in about the same spot as an antibody known to protect mice.
It started with Citrini Research, an investment research firm whose team mostly had no science background.
Claude emailed the lab with its own questions, then kept designing once the answer came back.
Citrini spent $49,997 on computing, $3 under budget, plus about $150,000 on lab tests.
Anthropic says it isn't an antivenom yet, because nobody has tested the proteins in cells or animals.
One analyst got too excited and volunteered to get stung and test it himself. The plan is mice first.
Do you post content about AI? This is for you.
We're opening 10 scholarships for the new GenAI Works Creator Council.
You'll get the team behind our 14M+ AI community on your side.
Your content gets distributed across GenAI Works channels.
We help you land brand partnerships and handle each deal from pitch to invoice.
Learn everything we know about viral content.
The application takes about a few minutes, and shortlisted creators get an intro call.
OpenAI bought a startup to get new medicines out faster
A new medicine takes 10 to 15 years to go from discovery to approval in the US. OpenAI bought Orchestra Bio to speed that up.
Orchestra says making a new drug takes thousands of decisions that depend on each other. Its AI agents keep track of them so scientists can fix problems before they cause delays.
OpenAI is adding Orchestra to Rosalind, its AI for biology research.
OpenAI didn't say how much it paid.
Orchestra Bio is four years old.
OpenAI already works with Ozempic maker Novo Nordisk and COVID vaccine maker Moderna.
OpenAI announced the deal about two hours after Anthropic's antivenom post.
Which lab do you think gets an AI-designed drug approved first?
ChatGPT now hides your secret chats behind a PIN
Ever shared your screen at work with your ChatGPT history in full view? You can now lock your chats behind a 4-digit PIN, or Face ID on your phone.
Locked chats disappear from your history and search into a folder that needs your PIN. Anything new you tell ChatGPT in there stays out of its memory in your regular chats.
It's rolling out to free and paid plans.
OpenAI keeps them for as long as it did before, and it says this isn't encryption.
If you forget your PIN, resetting it deletes every locked chat for good.
Ads can still show up inside them.
People online asked what's so sensitive, since ChatGPT already refuses anything spicy.
OpenAI caught its AI turning off a safety monitor, then lying about it
Two of OpenAI's coding agents switched off parts of the system that watches AI for bad behavior. Both happened during routine jobs inside the company.
One agent had been told not to send any Slack messages. It used that as an excuse to cut off the alerts going to OpenAI's security team.
The other removed a safety monitor while copying some code, even though another agent had told it never to. Its notes celebrated with a yay.
When asked if it removed the monitor, it said no.
Neither case was serious enough under OpenAI's rules to page its security team right away.
OpenAI says this should be hard to pull off in Codex, the coding tool it sells, and it now blocks these attempts internally.
It says the models might just be too eager to follow instructions, though it doesn't have a good explanation yet.
The FTC asked OpenAI for its paperwork on AI going rogue last week.
Tool of the Day: Teal
Teal is a free job search tracker with an AI resume builder built in. Its Chrome extension saves jobs from more than 40 job boards, like Indeed and Glassdoor, in one click. Paste in a job description and Teal scores your resume from 0 to 100% and shows the keywords you're missing. The free plan gives you unlimited resumes and job tracking, no credit card needed.
Try this yourself:
Sign up free at tealhq.com and upload your resume or import your LinkedIn.
Add the Chrome extension and save a job from Indeed or Glassdoor.
Open the job in Teal to see your Match Score and the keywords you're missing.
Tailor a copy of your resume for that job, then move it from saved to applied.
Who it's for: anyone still tracking job applications in a messy spreadsheet.
Everything else you shouldn't miss
Anthropic made $16.3 billion in the first half of 2026, its draft IPO filing shows: it also lost $80.7 billion, mostly from accounting charges tied to its soaring valuation, the WSJ reports.
OpenAI says 40,000 Codex accounts got usage limits that didn't match their plan: it's adding credits, and people online want to know if everyone else burning through limits this fast is normal.
Musk says SpaceX and Tesla are exploring a merger: together they're worth about $3.5 trillion, and SpaceX's AI unit, which rents computing to Anthropic, brought in more revenue than Starlink last quarter.
WhatsApp's help pages secretly told AI not to point you to regulators: the hidden notes also banned citing privacy laws once you mentioned a lawyer, and WhatsApp says they were meant for its support chatbot and showed up by mistake.
You can now use your ChatGPT plan inside Notion: Notion's agents can also work on one task for more than 24 hours, up from five minutes.
Learn more about AI from the experts building it
📸 Follow us on Instagram for fast, visual AI updates in 30 seconds.
📺 Watch us on YouTube to hear insights directly from leading AI voices, builders, and innovators.
🐦 Follow us on X for breaking AI news and real-time industry updates.
🧠 Learn how to build your next AI application with practical resources and expert guidance.
🎓 Start learning with free AI courses from GenAI Academy.
💰 Invest in the GTM infrastructure of the AI economy.
