Snippings

xxx

When you click a link inside your LLM or ask your agent to, the link now needs to carry a payload linking the AI citation to the merchant’s site, with consent signals (like a cookie, saying you consented to the agent doing this). In UCP it works like this:
The LLM looks for a file at [merchant.com/.well-known/ucp]. This manifest tells the LLM what the store is capable of (e.g., guest checkout, identity linking, or loyalty points).
The “link” carries a UCP Context Object. This tells the merchant site that the user didn’t just stumble upon the page — they were sent by an agent with a specific Intent.

From: (1) Simon Taylor on X: “The 5 Levels of Agentic Commerce” / X.

xxx

xxx

If you think about the 5 levels of autonomy in self-driving cars, Stripe had a similar model for commerce. These levels are typically: no automation, AI assistance, partial autonomy, full autonomy in certain conditions, high autonomy for some flows (or geographies), and finally full autonomy for any payment anywhere, any time.

From: (1) Simon Taylor on X: “The 5 Levels of Agentic Commerce” / X.

xxx

xxx

A threat actor took a “leisurely tour” of the city’s online resources and had started messing around with conference room projectors and other relatively harmless endpoints. Then they realized that they could change settings with the water utility where they switched many controls off, potentially endangering the water supply.

When Beckwith investigated, she found that all of the mischief was performed by an account that belonged to “Greg from Auditing.” There was just one problem. Greg hadn’t worked for the city for many years.
Unfortunately, even though Greg was no longer around, his account was, and it retained extensive privileges, including domain admin rights, SCADA (Supervisory Control and Data Acquisition) operator access, and even the ability to perform help desk functions. It’s unclear if someone from auditing ever needed this level of access, but a former employee definitely did not.

From: Zombie user account let hackers control the city’s water.

xxx

xxx

Millions of UK consumers are turning to social media influencers for financial guidance without checking their credentials, potentially putting their credit health at risk, according to new researchi commissioned by TransUnion.

The survey found that 14% of consumers, around 7.7 million peopleii, have taken financial advice from a social media personality or online influencer with a quarter (25%) of these consumers, approximately 1.9 million peopleiii, admitting they did not check whether the influencer had any formal financial qualifications or credentials before acting on the advice.

Among Gen Z consumers (aged 18-24), the use of financial influences rises sharply to 29% – roughly 1.4 million young peopleiv with almost a third (32%) of 18-24 year olds admitting they did not check the influencers qualifications before acting on the advice.

While some younger consumers reported benefits, the findings underline clear risks. Among 18-24 year olds who followed influencer advice, 39% said they gained useful financial knowledge and 31% said it helped them choose a good credit or financial product. However, 15% said following financial influencer advice negatively affected their credit score, led to financial losses or resulted in them being scammed.

From: Almost Two Million UK Consumers Take Financial Advice from Influencers Without Checking Their Credentials.

xxx

Here is an example: an impostor called Alight’s Benefits Information Center. She gave the mark’s name, the last four digits of her Social Security Number, her date of birth and a mailing address that matched what they had on file. That was enough to clear the security check and within a few weeks a check (yes, this was in America) for the full $750k from the account had been sent to an address in Las Vegas.

How can presenting publicly available information be considered adequate to pass a security check for a retirement savings account?

xxx

The article continues by noting that banks do not seem to be making as much of this interesting new technology as they might and that “what may prove to be more serious is the determination to cling to time-honoured procedures”. Well, yes indeed.

From: Editing “Ledgers and innovation in banking” – Substack.

xxx

xxx

Today at Google I/O 2026, we shared more about intelligent eyewear: glasses that deliver help in the moment without taking you out of it. There will be two types of intelligent eyewear: audio glasses that offer spoken help in your ear, and display glasses that show you the information you need, right when you need it. Both let you stay hands-free and heads up, and get you help from Gemini just by asking.

From: Intelligent eyewear with Gemini is coming this fall.

xxx

The Trump administration has just annouced that it is awarding $2 billion in grants to nine quantum-computing companies in deals that include U.S. government equity stakes, with $1 billion going to IBM. In other news, nearly one-third of all Bitcoin in circulation (worth over $469 billion at the time of writing) is vulnerable to theft if powerful quantum computers become a reality, according to research from blockchain analytics firm Glassnode.

xxx

“Perhaps there is an ulterior motive for putting those messages out; raising money or whatever,” Hassabis says. “From my point of view, from DeepMind and Google’s point of view, if engineers are becoming three or four times more productive, then we just [want to] do three or four times more stuff.”

From: Demis Hassabis Thinks AI Job Cuts Are Dumb | WIRED.

xxx

xxx

The double-spending problem is quite a problem. It is a general problem, not only about money, and fundamental to digital property. You have to go to inordinate lengths to make sure that if I send you some data on the Internet the data is now yours and not mine. It is a very difficult problem, much more complicated than sending pictures of cats, but the good news is that it is a solved problem.

From: (3) Sending Money is harder than Sending Pictures of Cats.

xxx