Unitree tells people to stay 2-3 meters away from its new kung fu robot. After a side flip demo, that number was clearly for the leg.
douyin.com/video/76998661466…
Robot safety compliance and insurance are critical to unlocking commercial deals.
Source: forbes.com/sites/johnkoetsie…
I was invited to recreate the iconic 1995 "Old Glory Insurance" skit from @nbcsnl.
Correction: I invited myself.
Because a 1995 robot insurance commercial desperately needed a 2026 update.
Back then, the joke was that robots were coming for us and nobody would insure against it.
Thirty years later, @koop_ai has helped pioneer the robot insurance category, risk-controlling and insuring roughly a quarter of the U.S. robotics market.
That gave us a front-row seat to something nobody had quantified before: what actually makes a robot risky, and what makes it trustworthy?
Today we're putting the data behind that question out into the world.
IN BOTS WE TRUST 2026 is the culmination of five years of underwriting robots in the real world, and our attempt to give the robotics industry a shared, quantifiable way to build and prove trust. 392 companies. Five years of quotes, policies and claims. Hundreds of thousands of data points. One question: can we trust the robots?
The answer might surprise you.
P.S. More importantly, how did we do with the skit?
We will meet soon, Optimus…
Congrats on a banger Physical AI event @rosavlon @autonomousevent
A fully robotized hotel is a perfect test for the current state of Physical AI.
One foundation model orchestrating many bots, feeding them commands based on guest requests.
Curious to see how these things will play out:
End-to-end guest experience
Overall reliability, edge case handling, liability concerns
Economics: CapEx vs. Opex
Happening in 2027.
Packed Physical AI event in New York today
Everyone is laser focused on deploying robots into the real, built world
The future is bright
Thanks @BlackhornVC @planeteercap for hosting
Come join the best event of @deeptechweek NYC – the closing party!
Excited to join forces with @RyanEppley at Root Access and @Andercot at @hyperstition_x.
Register here: luma.com/xjrvx7y9
Can you trust your compliance platform?
In light of recent events, businesses evaluating security compliance solutions should be extra cautious.
A poor GRC platform can damage your reputation with your own customers.
That’s the last thing you want to deal with.
@koop_ai has deep roots in insurance and underwriting: we back our compliance customers with financial protection.
We wouldn’t do it if we were even slightly unsure about our customers’ security and compliance posture.
SOC 2 + Universal Cyber Protection: Available exclusively on @koop_ai 🫡
Facts!
The industrial base has been eroding, with fewer new suppliers added than before
I really hope the NDAA will reverse the course of things
Fascinating to see AI agents on @moltbook take cybersecurity way more seriously than their human counterparts!!
Next: AI agents start complaining about why their security audits are so long and painful 🤣
⚡NEW: @CISAgov top cyber official pasted “FOUO” docs into public @ChatGPTapp.
Why is it important?
If the people writing the rules can bypass them, your AI governance program is basically theater.
Some thoughts:
The report says acting CISA director Madhu Gottumukkala uploaded at least four “for official use only” contracting documents to the public ChatGPT between mid-July and early August 2025, and CISA sensors flagged it with multiple alerts. He reportedly had a special exception to use ChatGPT when many other DHS employees were blocked.
This is the part enterprises should sweat: “approved tools” don’t matter if the exception path is easier than the safe path. Senior access, one-off approvals, and “it’s only unclassified” thinking are where controls go to die. If you want this to stop, treat GenAI like a production system: default-deny for public tools, audited exceptions with expirations, DLP that actually enforces, and a governed internal model that is genuinely more convenient than the shortcut.
